spg_engine/eval.rs
1//! Expression evaluator. Given a parsed `Expr`, a `Row`, and the row's column
2//! schema, produce a `Value`. v0.4 implements:
3//!
4//! - literals
5//! - column lookups (bare and qualified `t.col`)
6//! - unary minus / NOT
7//! - binary arithmetic, comparison, AND, OR
8//! - numeric widening (`Int → BigInt → Float`) at evaluation time
9//! - SQL three-valued logic for NULL:
10//! * any arithmetic / comparison op with a NULL operand → NULL
11//! * `TRUE OR NULL` → TRUE, `FALSE OR NULL` → NULL,
12//! * `FALSE AND NULL` → FALSE, `TRUE AND NULL` → NULL,
13//! * `NOT NULL` → NULL
14//!
15//! v0.4 deliberately does *not* implement: function calls, string
16//! concatenation, IS NULL / IS NOT NULL, BETWEEN, IN, etc. Those come later.
17
18use alloc::borrow::Cow;
19use alloc::format;
20use alloc::string::{String, ToString};
21use alloc::vec::Vec;
22
23use spg_sql::ast::{BinOp, CastTarget, ColumnName, Expr, Literal};
24use spg_storage::{ColumnSchema, Row, Value};
25
26pub(crate) mod binop;
27mod cast;
28pub mod compiled;
29mod datetime;
30mod encoding;
31mod encodings;
32mod format;
33pub(crate) mod functions;
34mod inet;
35pub(crate) mod math;
36mod regexp;
37mod resolve;
38mod strings;
39mod textsearch;
40pub(crate) mod values;
41
42pub use crate::conversions::format_money_array;
43pub(crate) use binop::{
44 add_interval_to_micros, and_3vl, apply_binary, apply_binary_by_ref, apply_binary_interval,
45};
46use binop::{apply_binary_in, apply_unary, compare, pow10_i128};
47pub use cast::{cast_to_vector, cast_value, parse_vector_text};
48pub(crate) use compiled::{
49 CompiledExpr, compile_column_pos, compile_expr, eval_compiled, eval_compiled_ref,
50 fully_compilable,
51};
52use datetime::{
53 age, date_format_mysql, date_part, date_trunc, extract_field, from_unixtime, unix_timestamp_of,
54};
55use encoding::{decode_text, encode_text};
56pub use format::{
57 days_from_civil, format_bigint_array, format_bool_array, format_bytea_array, format_bytea_hex,
58 format_date, format_date_array, format_float, format_float_array, format_int_array,
59 format_interval, format_interval_array, format_money, format_numeric, format_numeric_array,
60 format_numeric_kind, format_real, format_smallint_array, format_text_array, format_time,
61 format_timestamp, format_timestamp_array, format_timestamptz, format_timestamptz_at,
62 format_timetz, format_uuid_array, parse_date_literal, parse_timestamp_literal,
63};
64// v7.39 (GUC knife 3) — session render styles + styled formatters.
65pub use format::{
66 DateOrder, DateStyleKind, IntervalStyleKind, RenderStyle, format_date_array_styled,
67 format_date_styled, format_float_array_styled, format_float_styled,
68 format_interval_array_styled, format_interval_styled, format_real_styled,
69 format_timestamp_array_styled, format_timestamp_styled, format_timestamptz_styled,
70 format_timestamptz_tz, parse_date_literal_ordered, parse_timestamp_literal_ordered,
71 parse_timestamp_literal_tz_ordered,
72};
73use functions::apply_function;
74use inet::{inet_host, inet_masklen, inet_network, inet_op_bool_result};
75pub(crate) use math::{f64_ceil, f64_floor, f64_sqrt};
76use math::{
77 f64_exp, f64_ln, f64_powi, f64_round_half_away, f64_trunc, prng_next_f64, prng_next_u64,
78};
79pub(crate) use regexp::{
80 CompiledRe, compile_re, compiled_is_match, regex_is_match, regexp_matches_rows,
81};
82use regexp::{regexp_matches, regexp_replace, regexp_split_to_array};
83use resolve::{
84 collation_fold_for_compare, compare_is_case_insensitive, composite_eq, eval_expr_cow,
85 is_owned_compare_value, resolve_column, resolve_column_borrowed, text_prefix_chars,
86};
87pub(crate) use resolve::{
88 column_at, column_collation, find_column_pos, is_binary_coerced, locate_column,
89};
90use strings::{
91 TrimSide, format_string, pg_quote_ident, pg_quote_literal, pg_typeof_name, string_left_right,
92 string_pad, string_trim, to_char, value_to_format_text,
93};
94pub use textsearch::{
95 decode_tsquery_external, decode_tsvector_external, format_tsquery, format_tsvector,
96};
97use textsearch::{
98 fts_phraseto_tsquery, fts_plainto_tsquery, fts_setweight, fts_to_tsquery, fts_to_tsvector,
99 fts_ts_headline, fts_ts_rank, fts_ts_rank_cd, fts_ts_rewrite, fts_tsquery_bool,
100 fts_websearch_to_tsquery, ts_match, tsvector_concat,
101};
102pub use values::gen_random_uuid_bytes;
103/// v7.39 (tz epic) — fixed-offset / abbreviation resolution, exposed
104/// for `SET timezone` validation (named zones go through the host tzdb).
105pub(crate) fn datetime_resolve_zone_offset(z: &str) -> Option<i64> {
106 datetime::resolve_zone_offset(z)
107}
108
109pub use values::value_to_text;
110pub use values::value_to_text_styled;
111pub use values::value_to_text_with_fsp;
112use values::{
113 array_2d_dims, array_element_at, array_len, array_rebuild, value_cmp_for_min_max, value_to_f64,
114 values_equal_for_nullif,
115};
116
117/// Resolution context for evaluating a single row. `table_alias` is the alias
118/// (or table name) callers should accept as the qualifier on a column ref —
119/// e.g. `FROM users AS u` makes `u.name` valid and rejects `other.name`.
120#[derive(Clone)]
121#[allow(missing_debug_implementations)] // sequence_resolver is a dyn Fn — no Debug
122pub struct EvalContext<'a> {
123 pub columns: &'a [ColumnSchema],
124 pub table_alias: Option<&'a str>,
125 /// v6.1.1 — bound parameters for `$N` placeholders inside the
126 /// expression tree. Empty for simple queries; populated by the
127 /// prepared-statement Execute path with Bind values converted
128 /// to `Value`. Index N (1-based per PG) hits `params[N-1]`.
129 pub params: &'a [Value<'static>],
130 /// v7.12.1 — session text-search config (from `SET
131 /// default_text_search_config = '<name>'`). Resolved when the
132 /// engine builds an `EvalContext` and consumed by the FTS
133 /// function dispatcher when `to_tsvector(text)` /
134 /// `plainto_tsquery(text)` etc are called without an explicit
135 /// config arg. `None` falls through to `simple`.
136 pub default_text_search_config: Option<&'a str>,
137 /// v7.17.0 Phase 1.1 — `nextval` / `currval` / `setval`
138 /// resolver. The engine builds this around a `&mut Catalog`
139 /// so apply_function can mutate sequence state without
140 /// eval owning a catalog reference. When `None`, sequence
141 /// functions return an error (read-only contexts).
142 pub sequence_resolver: Option<&'a SequenceResolver<'a>>,
143 /// v7.37.16 (16.12) — read-only catalog reference for
144 /// builtins that need catalog walks (e.g. `pg_partition_root`,
145 /// `pg_partition_ancestors`). `None` falls through to the
146 /// "no catalog available" branch which returns NULL — same
147 /// shape PG returns for a non-existent OID. Most evaluation
148 /// sites don't need catalog access (row scans, projections);
149 /// they construct contexts with `catalog: None` and the
150 /// engine populates `Some(&self.catalog)` only at the engine's
151 /// top-level entry points where the borrow is unambiguous.
152 pub catalog: Option<&'a spg_storage::Catalog>,
153 /// v7.39 (round 346, M1) — is this a MySQL-dialect session? The two
154 /// dialects disagree about what counts as a truth value: MariaDB
155 /// takes any non-zero number (and a string's leading number) as
156 /// true, PG refuses anything that is not boolean. Set from the
157 /// engine by [`EvalContext::with_engine`]; a context built without
158 /// one keeps PG's stricter reading.
159 pub mysql_dialect: bool,
160 /// Session GUCs set via `SET name = value` / `set_config`, keyed by
161 /// lowercased name. `current_setting('app.foo')` reads custom
162 /// (namespaced) settings from here — the mechanism apps use for
163 /// request context / RLS. `None` in read-only contexts that have no
164 /// session; unknown names then fall through to PG defaults.
165 pub session_gucs: Option<&'a alloc::collections::BTreeMap<String, String>>,
166 /// v7.39 (read01 round 58) — the engine's role store, so
167 /// `has_table_privilege('bob', …)` can expand bob's role MEMBERSHIPS (a
168 /// grant to a group role answers `true` for its inheriting members). `None`
169 /// in a context with no engine behind it — the role then stands alone.
170 pub users: Option<&'a crate::users::UserStore>,
171 /// v7.39 (read01 round 61) — how deep we are inside USER-DEFINED function
172 /// bodies. A function's body is evaluated with a child context, and a body
173 /// may call another function, so this bounds the recursion (a function that
174 /// calls itself would otherwise blow the stack, which an embed host cannot
175 /// catch).
176 pub fn_depth: u16,
177 /// v7.39 (read01 round 63) — the ENGINE, for a user-function body that has
178 /// its own FROM (`SELECT v FROM t WHERE id = k`). Such a body has to run
179 /// through the real executor: reading `catalog`'s rows straight from eval
180 /// would bypass the row-header visibility filter, so under in-place MVCC a
181 /// function would happily read DEAD rows. `None` in a context with no
182 /// engine behind it — a body with a FROM then errors, saying so.
183 pub engine: Option<&'a crate::Engine>,
184 /// v7.38 (read01 U15) — per-scan deterministic sampler state for
185 /// `TABLESAMPLE … REPEATABLE(seed)`. A fresh cell is created before a
186 /// scan whose predicate may draw `__tsm_fract(seed)`; the cell holds
187 /// `None` until the first draw seeds it from that literal, then a
188 /// scan-local xorshift sequence (isolated from the process-global
189 /// `random()` PRNG, so it's deterministic and rescan-stable). `None`
190 /// here means no sampler is attached.
191 pub sample_rng: Option<&'a core::cell::Cell<Option<u64>>>,
192 /// v7.38 (read01 P3.25) — native-stack-overflow guard. Lazily seeded
193 /// with the stack pointer of the outermost `eval_expr` call; deeper
194 /// calls compare their own pointer against it and bail with
195 /// [`EvalError::StackDepthExceeded`] once usage crosses a safe margin,
196 /// so a pathologically nested expression errors instead of aborting the
197 /// process. Owned (not a borrowed cell) so it stays stack-local and
198 /// never touches `Engine`'s `Sync` bound.
199 pub recursion_base: core::cell::Cell<usize>,
200 /// v7.39 (GUC knife 3) — session render style (DateStyle /
201 /// IntervalStyle / extra_float_digits) for text output produced
202 /// inside expression evaluation (`::text` casts). Contexts built
203 /// away from the session (per-shard scan filters, index probes)
204 /// keep the default — they don't render text output.
205 pub render_style: crate::eval::format::RenderStyle,
206 /// v7.39 (tz epic) — host IANA timezone lookups for named zones
207 /// (session rendering, AT TIME ZONE, literal zone suffixes).
208 pub tz_offset_fn: Option<crate::TzOffsetFn>,
209 pub tz_localize_fn: Option<crate::TzLocalizeFn>,
210 pub tz_abbrev_fn: Option<crate::TzAbbrevFn>,
211 /// v7.38 (read01 P5.24) — host-provided CSPRNG (the server injects
212 /// `/dev/urandom`). Cryptographic builtins (`gen_random_bytes`,
213 /// `gen_salt`) draw from this instead of the process-static xorshift
214 /// PRNG, so their output isn't predictable. `None` (no host CSPRNG)
215 /// falls back to the PRNG — fine for the non-cryptographic `random()`.
216 pub salt_fn: Option<crate::SaltFn>,
217 /// v7.39 (read01 pgstatfuncs.c) — calling-connection identity for
218 /// pg_backend_pid(); `None` (embedded / detached contexts) → 1.
219 pub backend_pid_fn: Option<crate::BackendPidFn>,
220 /// v7.39 (round 476) — the WAL byte position, for the LSN functions.
221 pub wal_lsn_fn: Option<crate::WalLsnFn>,
222 /// v7.39 (round 318, V51) — host connection-control hook for
223 /// `pg_cancel_backend` / `pg_terminate_backend`. `None` (embedded /
224 /// detached contexts) ⇒ there is nothing to signal, so they answer
225 /// false rather than pretending the signal landed.
226 pub backend_signal_fn: Option<crate::BackendSignalFn>,
227 /// v7.38 (read01 P6.08) — host wall clock (µs since Unix epoch). `uuidv7`
228 /// uses it for the real time-ordered 48-bit millisecond prefix; `None`
229 /// (no host clock) falls back to the deterministic anchor.
230 pub clock: Option<crate::ClockFn>,
231 /// v7.38 (T24) — read-only view of the engine's transaction-version state,
232 /// so the `txid_*` / `pg_*_xact_id` / `pg_xact_status` builtins report the
233 /// real transaction ids instead of a constant stub. `None` on the scan /
234 /// join / aggregate contexts that never evaluate them.
235 pub xact: Option<XactView<'a>>,
236 /// v7.38 (T24) — PG's `txid_current()` ASSIGNS an id to a transaction that
237 /// has none. In autocommit a read-only statement has no writer version, so
238 /// the first call allocates one here and later calls in the same statement
239 /// reuse it — `SELECT txid_current(), txid_current()` must agree, as in PG.
240 pub assigned_xid: core::cell::Cell<Option<u64>>,
241}
242
243/// v7.38 (T24) — the transaction-id surface PG's `txid_*` family exposes.
244/// SPG's writer versions ARE its transaction ids (`row_header::next_version`),
245/// so no separate xid counter is needed — this is the bridge U22 was waiting
246/// on.
247#[derive(Clone, Copy, Debug)]
248pub struct XactView<'a> {
249 /// The id assigned to the current transaction (allocated at BEGIN) or, in
250 /// autocommit, to the current statement once it has written. `None` when
251 /// nothing has been assigned — `*_if_assigned` returns NULL there, as PG does.
252 pub current: Option<u64>,
253 /// Ids allocated by transactions that have neither committed nor aborted.
254 pub active: &'a alloc::collections::BTreeSet<u64>,
255 /// Ids of rolled-back transactions.
256 pub aborted: &'a alloc::collections::BTreeSet<u64>,
257}
258
259/// v7.17.0 — sequence-mutating callback used by `apply_function`
260/// for `nextval` / `currval` / `setval`. Implemented by the
261/// engine to thread `&mut Catalog` access through an immutable
262/// `&EvalContext`.
263pub type SequenceResolver<'a> = dyn Fn(SequenceOp) -> Result<i64, EvalError> + 'a;
264
265/// v7.17.0 — sequence operation requested by an Expr eval.
266#[derive(Debug, Clone)]
267pub enum SequenceOp {
268 Next(String),
269 Curr(String),
270 Set {
271 name: String,
272 value: i64,
273 is_called: bool,
274 },
275}
276
277impl<'a> EvalContext<'a> {
278 pub const fn new(columns: &'a [ColumnSchema], table_alias: Option<&'a str>) -> Self {
279 Self {
280 columns,
281 table_alias,
282 params: &[],
283 default_text_search_config: None,
284 sequence_resolver: None,
285 catalog: None,
286 mysql_dialect: false,
287 session_gucs: None,
288 users: None,
289 fn_depth: 0,
290 engine: None,
291 sample_rng: None,
292 recursion_base: core::cell::Cell::new(0),
293 render_style: crate::eval::format::RenderStyle {
294 date_style: crate::eval::format::DateStyleKind::Iso,
295 date_order: crate::eval::format::DateOrder::Mdy,
296 interval_style: crate::eval::format::IntervalStyleKind::Postgres,
297 extra_float_digits: 1,
298 bytea_escape: false,
299 mysql: false,
300 },
301 tz_offset_fn: None,
302 tz_localize_fn: None,
303 tz_abbrev_fn: None,
304 salt_fn: None,
305 backend_pid_fn: None,
306 wal_lsn_fn: None,
307 backend_signal_fn: None,
308 clock: None,
309 xact: None,
310 assigned_xid: core::cell::Cell::new(None),
311 }
312 }
313
314 /// v7.39 (GUC knife 3) — attach the session render style.
315 #[must_use]
316 pub const fn with_render_style(mut self, style: crate::eval::format::RenderStyle) -> Self {
317 self.render_style = style;
318 self
319 }
320
321 /// v7.39 (round 318, V51) — attach the host connection-control hook.
322 #[must_use]
323 pub const fn with_backend_signal_fn(mut self, f: Option<crate::BackendSignalFn>) -> Self {
324 self.backend_signal_fn = f;
325 self
326 }
327
328 /// v7.39 (round 476) — attach the WAL byte-position provider.
329 #[must_use]
330 pub const fn with_wal_lsn_fn(mut self, f: Option<crate::WalLsnFn>) -> Self {
331 self.wal_lsn_fn = f;
332 self
333 }
334
335 /// v7.39 (read01 pgstatfuncs.c) — attach the calling-connection id.
336 #[must_use]
337 pub const fn with_backend_pid_fn(mut self, f: Option<crate::BackendPidFn>) -> Self {
338 self.backend_pid_fn = f;
339 self
340 }
341
342 /// v7.39 (tz epic) — attach the host timezone lookups.
343 #[must_use]
344 pub const fn with_tz_fns(
345 mut self,
346 offset: Option<crate::TzOffsetFn>,
347 localize: Option<crate::TzLocalizeFn>,
348 abbrev: Option<crate::TzAbbrevFn>,
349 ) -> Self {
350 self.tz_offset_fn = offset;
351 self.tz_localize_fn = localize;
352 self.tz_abbrev_fn = abbrev;
353 self
354 }
355
356 /// v7.39 (tz epic) — offset (µs east) of an arbitrary zone spec at
357 /// a UTC instant: fixed forms resolve statically, named zones
358 /// through the host tzdb. None = unknown zone.
359 #[must_use]
360 pub fn zone_offset_at(&self, zone: &str, utc_micros: i64) -> Option<i64> {
361 if let Some(off) = datetime::resolve_zone_offset(zone) {
362 return Some(off);
363 }
364 self.tz_offset_fn.and_then(|f| f(zone, utc_micros))
365 }
366
367 /// v7.39 (tz epic) — the SESSION zone's offset at a UTC instant
368 /// (per-value: DST zones vary within one statement).
369 #[must_use]
370 pub fn session_tz_offset_at(&self, utc_micros: i64) -> i64 {
371 let Some(zone) = self.session_gucs.and_then(|g| g.get("timezone")) else {
372 return 0;
373 };
374 self.zone_offset_at(zone, utc_micros).unwrap_or(0)
375 }
376
377 /// v7.39 (tz epic) — the session zone's designation at an instant
378 /// (named zones only; None lets renderers spell UTC / +HH).
379 #[must_use]
380 pub fn session_tz_abbrev_at(&self, utc_micros: i64) -> Option<alloc::string::String> {
381 let zone = self.session_gucs.and_then(|g| g.get("timezone"))?;
382 if datetime::resolve_zone_offset(zone).is_some()
383 || zone.eq_ignore_ascii_case("utc")
384 || zone.eq_ignore_ascii_case("gmt")
385 {
386 return None;
387 }
388 self.tz_abbrev_fn.and_then(|f| f(zone, utc_micros))
389 }
390
391 /// v7.39 (tz epic) — local wall micros in `zone` -> UTC micros
392 /// (PG's DST disambiguation for named zones).
393 #[must_use]
394 pub fn zone_local_to_utc(&self, zone: &str, local_micros: i64) -> Option<i64> {
395 zone_local_to_utc_with(zone, local_micros, self.tz_localize_fn)
396 }
397
398 /// v7.38 (read01 P5.24) — attach the host CSPRNG so cryptographic
399 /// builtins don't fall back to the predictable PRNG.
400 #[must_use]
401 pub const fn with_salt_fn(mut self, f: Option<crate::SaltFn>) -> Self {
402 self.salt_fn = f;
403 self
404 }
405
406 /// v7.38 (read01 P6.08) — attach the host wall clock so `uuidv7` gets a
407 /// real time-ordered prefix instead of the deterministic anchor.
408 #[must_use]
409 pub const fn with_clock(mut self, f: Option<crate::ClockFn>) -> Self {
410 self.clock = f;
411 self
412 }
413
414 /// v7.38 (read01 U15) — attach a per-scan `TABLESAMPLE REPEATABLE`
415 /// sampler cell. The cell (seeded lazily on first `__tsm_fract` draw)
416 /// must outlive the context and be created fresh per scan so a rescan
417 /// re-seeds and reproduces the same sample.
418 #[must_use]
419 pub const fn with_sample_rng(mut self, cell: &'a core::cell::Cell<Option<u64>>) -> Self {
420 self.sample_rng = Some(cell);
421 self
422 }
423
424 /// Attach the session's GUC map so `current_setting` can resolve
425 /// custom (namespaced) settings written with `SET` / `set_config`.
426 #[must_use]
427 /// v7.39 (read01 round 63) — thread the engine (see `engine`).
428 pub const fn with_engine(mut self, engine: &'a crate::Engine) -> Self {
429 self.mysql_dialect = engine.backslash_escapes;
430 // v7.39 (round 368, M20 P3) — the dialect also decides how a binary
431 // string renders in a string context (latin-1 bytes vs PG `\x…`).
432 self.render_style.mysql = engine.backslash_escapes;
433 self.engine = Some(engine);
434 self
435 }
436
437 /// v7.39 (read01 round 58) — thread the role store (see `users`).
438 pub const fn with_users(mut self, users: &'a crate::users::UserStore) -> Self {
439 self.users = Some(users);
440 self
441 }
442
443 /// v7.39 (round 524) — attach a whole session bag at once. Every
444 /// write path needs the same four, and taking them one at a time is
445 /// how three of them ended up with none.
446 #[must_use]
447 pub(crate) fn with_session<'b: 'a>(mut self, s: &'b DmlSession) -> Self {
448 self.session_gucs = Some(&s.gucs);
449 self.users = Some(&s.users);
450 self.render_style = s.render_style;
451 self.tz_offset_fn = s.tz_offset_fn;
452 self.tz_localize_fn = s.tz_localize_fn;
453 self.tz_abbrev_fn = s.tz_abbrev_fn;
454 self
455 }
456
457 pub const fn with_session_gucs(
458 mut self,
459 gucs: &'a alloc::collections::BTreeMap<String, String>,
460 ) -> Self {
461 self.session_gucs = Some(gucs);
462 self
463 }
464
465 /// v7.37.16 (16.12) — attach a read-only catalog reference
466 /// so builtins like `pg_partition_root` can walk partition
467 /// roles. Defaults to None (NULL semantics).
468 #[must_use]
469 pub const fn with_catalog(mut self, catalog: &'a spg_storage::Catalog) -> Self {
470 self.catalog = Some(catalog);
471 self
472 }
473
474 /// v7.38 (T-tstz Phase 2) — the micro-offset of the session `TimeZone` GUC
475 /// (`SET TimeZone = '+09'` → +9h). A fixed offset / abbreviation resolves;
476 /// UTC and an unset GUC give 0; a named IANA zone (no tzdata) also gives 0
477 /// so a timestamptz still renders — as `+00` — rather than erroring on
478 /// every display. Timestamptz rendering / cast is the only consumer.
479 #[must_use]
480 pub fn session_tz_offset(&self) -> i64 {
481 self.session_gucs
482 .and_then(|g| g.get("timezone"))
483 .and_then(|z| datetime::resolve_zone_offset(z))
484 .unwrap_or(0)
485 }
486
487 /// v7.38 (T24) — attach the transaction-version view the `txid_*` builtins
488 /// read. Defaults to None, where they fall back to the process-wide cursor.
489 #[must_use]
490 pub const fn with_xact(mut self, xact: XactView<'a>) -> Self {
491 self.xact = Some(xact);
492 self
493 }
494
495 /// v7.17.0 — attach a sequence resolver. The engine wraps a
496 /// `&mut Catalog` in a closure that performs the requested
497 /// SequenceOp.
498 #[must_use]
499 pub const fn with_sequence_resolver(mut self, resolver: &'a SequenceResolver<'a>) -> Self {
500 self.sequence_resolver = Some(resolver);
501 self
502 }
503
504 /// v6.1.1 — attach a parameter buffer for `$N` placeholder
505 /// resolution. The slice must outlive the context; callers
506 /// construct it from the prepared statement's Bind values.
507 #[must_use]
508 pub const fn with_params(mut self, params: &'a [Value<'static>]) -> Self {
509 self.params = params;
510 self
511 }
512
513 /// v7.12.1 — attach the session's
514 /// `default_text_search_config`. Used by the FTS function
515 /// dispatcher when no explicit config arg is given.
516 #[must_use]
517 pub const fn with_default_text_search_config(mut self, cfg: Option<&'a str>) -> Self {
518 self.default_text_search_config = cfg;
519 self
520 }
521}
522
523/// v7.39 (round 523) — read a timestamp literal, reporting whether it
524/// carried an offset. Re-exported for the INSERT path, which decides
525/// there whether a value already names an instant.
526pub(crate) fn parse_timestamp_literal_tz_ordered_pub(
527 s: &str,
528 order: DateOrder,
529) -> Option<(i64, bool)> {
530 format::parse_timestamp_literal_tz_ordered(s, order)
531}
532
533/// v7.39 (round 523) — a FIXED zone's offset, when the name is one
534/// (`+09`, `UTC-5`). Named zones go through the host's tzdb instead.
535#[must_use]
536pub(crate) fn resolve_zone_offset_pub(zone: &str) -> Option<i64> {
537 datetime::resolve_zone_offset(zone)
538}
539
540/// v7.39 (round 523) — a wall-clock reading in `zone` as a UTC instant.
541///
542/// A free function because the INSERT path needs it too, and that path
543/// carries no `EvalContext`: it evaluates VALUES through a context-free
544/// literal walker. `EvalContext::zone_local_to_utc` delegates here so the
545/// two cannot drift.
546#[must_use]
547pub(crate) fn zone_local_to_utc_with(
548 zone: &str,
549 local_micros: i64,
550 localize: Option<crate::TzLocalizeFn>,
551) -> Option<i64> {
552 if let Some(off) = datetime::resolve_zone_offset(zone) {
553 return Some(local_micros - off);
554 }
555 localize.and_then(|f| f(zone, local_micros))
556}
557
558/// v7.39 (round 523) — the session zone an assignment to a timestamptz
559/// column is read in, or `None` when the session is on UTC and no shift
560/// applies.
561#[derive(Debug, Clone)]
562pub(crate) struct SessionCoercion {
563 /// The session zone, when it is not UTC. `None` leaves an instant
564 /// where it was.
565 pub zone: Option<alloc::string::String>,
566 pub localize: Option<crate::TzLocalizeFn>,
567 /// The session's date order. A written date is ambiguous
568 /// (`01/02/2020`), and this is what resolves it.
569 pub order: DateOrder,
570}
571
572impl SessionCoercion {
573 /// The UTC instant a naive wall-clock reading names in the session
574 /// zone, or `None` when the session is on UTC.
575 #[must_use]
576 pub(crate) fn wall_to_utc(&self, wall: i64) -> Option<i64> {
577 let zone = self.zone.as_ref()?;
578 zone_local_to_utc_with(zone, wall, self.localize)
579 }
580
581 /// v7.39 (round 524) — the session facts an ASSIGNMENT is read
582 /// under, from an evaluation context. `None` when both are the
583 /// defaults and nothing needs re-reading.
584 #[must_use]
585 pub(crate) fn from_ctx(ctx: &EvalContext<'_>) -> Option<Self> {
586 let zone = ctx
587 .session_gucs
588 .and_then(|g| g.get("timezone"))
589 .filter(|z| !z.eq_ignore_ascii_case("utc") && !z.eq_ignore_ascii_case("gmt"))
590 .cloned();
591 let order = ctx.render_style.date_order;
592 if zone.is_none() && order == DateOrder::Mdy {
593 return None;
594 }
595 Some(Self {
596 zone,
597 localize: ctx.tz_localize_fn,
598 order,
599 })
600 }
601}
602
603/// v7.39 (round 524) — the session facts a DML evaluation context needs,
604/// cloned so the row loop can still borrow the engine mutably.
605///
606/// Every write path built a BARE `EvalContext`, so an expression in an
607/// UPDATE's SET or a DELETE's WHERE was evaluated by an engine that knew
608/// nothing about the connection. One value, built once per statement,
609/// and a grep for `dml_session` finds every path that has it.
610pub(crate) struct DmlSession {
611 pub gucs: alloc::collections::BTreeMap<String, String>,
612 pub users: crate::users::UserStore,
613 pub render_style: RenderStyle,
614 pub tz_offset_fn: Option<crate::TzOffsetFn>,
615 pub tz_localize_fn: Option<crate::TzLocalizeFn>,
616 pub tz_abbrev_fn: Option<crate::TzAbbrevFn>,
617}
618
619/// v7.39 (round 524) — read a TEXT value bound for a temporal column
620/// under the session's date order.
621///
622/// `01/02/2020` is February 1st in a DMY session and January 2nd in an
623/// MDY one, and the write path was reading every one of them as MDY: a
624/// `SELECT '01/02/2020'::date` answered PG's value while the same
625/// literal INSERTed stored the day and month swapped. Nothing errors,
626/// and once stored the two readings are indistinguishable.
627#[must_use]
628pub(crate) fn session_read_temporal_text(
629 v: Value<'static>,
630 target: spg_storage::DataType,
631 coercion: Option<&SessionCoercion>,
632) -> Value<'static> {
633 use spg_storage::DataType as D;
634 let Some(c) = coercion else { return v };
635 if c.order == DateOrder::Mdy {
636 return v;
637 }
638 let Value::Text(s) = &v else { return v };
639 match target {
640 D::Date => format::parse_date_literal_ordered(s, c.order).map_or(v, Value::Date),
641 D::Timestamp | D::Timestamptz => format::parse_timestamp_literal_tz_ordered(s, c.order)
642 .map_or(v, |(t, _)| Value::Timestamp(t)),
643 _ => v,
644 }
645}
646
647#[derive(Debug, Clone, PartialEq)]
648pub enum EvalError {
649 ColumnNotFound {
650 name: String,
651 },
652 UnknownQualifier {
653 qualifier: String,
654 },
655 DivisionByZero,
656 TypeMismatch {
657 detail: String,
658 },
659 /// v6.1.1 — `$N` reference past the number of bound parameters.
660 /// Either the client sent too few in Bind, or the SQL has a
661 /// placeholder the prepared statement didn't account for.
662 PlaceholderOutOfRange {
663 n: u16,
664 bound: u16,
665 },
666 /// v7.38 (read01 P3.25) — the expression tree recursed deep enough to
667 /// threaten a native stack overflow; we bail out with an error the way
668 /// PG's `check_stack_depth()` does instead of aborting the process.
669 StackDepthExceeded,
670}
671
672impl core::fmt::Display for EvalError {
673 fn fmt(&self, f: &mut core::fmt::Formatter<'_>) -> core::fmt::Result {
674 match self {
675 // v7.39 (read01 round 81) — PG's wording (and SQLSTATE trigger):
676 // `column "x" does not exist`, 42703. The old "column not found: x"
677 // matched none of the wire layer's `does not exist` patterns, so a
678 // missing column reached the client as the generic error class.
679 Self::ColumnNotFound { name } => write!(f, "column \"{name}\" does not exist"),
680 // v7.39 (round 241) — PG's wording (and 42P01 trigger): a
681 // qualifier that names no table in scope is "missing
682 // FROM-clause entry for table \"x\"". The old "unknown table
683 // qualifier" matched nothing a driver branches on.
684 Self::UnknownQualifier { qualifier } => {
685 write!(f, "missing FROM-clause entry for table \"{qualifier}\"")
686 }
687 Self::DivisionByZero => f.write_str("division by zero"),
688 Self::TypeMismatch { detail } => write!(f, "type mismatch: {detail}"),
689 Self::PlaceholderOutOfRange { n, bound } => write!(
690 f,
691 "parameter ${n} referenced but only {bound} bound by client"
692 ),
693 Self::StackDepthExceeded => {
694 f.write_str("stack depth limit exceeded (expression nested too deeply)")
695 }
696 }
697 }
698}
699
700/// v7.38 (read01 P3.25) — native-stack budget below the outermost
701/// `eval_expr` frame. Native stacks are typically 2–8 MB; 768 KiB leaves
702/// generous headroom while still permitting PG-class nesting depth (in a
703/// release build ~hundreds-to-thousands of frames fit under this).
704const MAX_EVAL_STACK_BYTES: usize = 768 * 1024;
705
706/// Address of a local in the current frame — a portable stand-in for the
707/// stack pointer (stacks grow downward on all supported targets).
708#[inline(never)]
709fn eval_stack_ptr() -> usize {
710 let probe = 0u8;
711 core::ptr::addr_of!(probe) as usize
712}
713
714/// v7.38 (read01 P6.40) — enforce a user DOMAIN's NOT NULL + CHECK constraints
715/// on a value being cast to it (`x::domain`). NULL fails a NOT NULL domain;
716/// otherwise every CHECK (which references the pseudo-column `VALUE`) must not
717/// evaluate to false. Returns the value unchanged when all constraints pass.
718fn apply_domain_constraints<'a>(
719 v: Value<'a>,
720 dom: &spg_storage::DomainDef,
721 name: &str,
722 cat: &spg_storage::Catalog,
723) -> Result<Value<'a>, EvalError> {
724 if matches!(v, Value::Null) {
725 // A NOT NULL anywhere in the chain rejects a NULL.
726 let mut cur = Some(dom);
727 while let Some(d) = cur {
728 if !d.nullable {
729 return Err(EvalError::TypeMismatch {
730 detail: alloc::format!("domain {name} does not allow null values"),
731 });
732 }
733 cur = d
734 .base_domain
735 .as_ref()
736 .and_then(|p| cat.domain_types().get(p.as_str()));
737 }
738 return Ok(v);
739 }
740 // v7.39 (round 259) — walk the domain chain BASE-FIRST (probed: a
741 // value violating both a parent's and the child's constraint reports
742 // the PARENT's). The message names the domain being cast TO, but the
743 // constraint that actually failed — `value for domain pchild violates
744 // check constraint "pbase_check"`.
745 let mut chain: alloc::vec::Vec<&spg_storage::DomainDef> = alloc::vec![dom];
746 let mut cur = dom;
747 while let Some(parent) = cur
748 .base_domain
749 .as_ref()
750 .and_then(|p| cat.domain_types().get(p.as_str()))
751 {
752 // A cycle cannot be created through CREATE DOMAIN (the parent must
753 // already exist), but stop defensively rather than loop forever.
754 if chain.iter().any(|d| core::ptr::eq(*d, parent)) {
755 break;
756 }
757 chain.push(parent);
758 cur = parent;
759 }
760 chain.reverse();
761 for owner in chain {
762 apply_domain_checks_of(&v, owner, name)?;
763 }
764 Ok(v)
765}
766
767/// v7.39 (round 259) — run ONE domain's own CHECK list against `v`. The
768/// error names `target` (the domain the value is being cast to) and
769/// `owner` (whose constraint failed); for a single-level domain they are
770/// the same, which is the pre-259 wording.
771fn apply_domain_checks_of(
772 v: &Value<'_>,
773 dom: &spg_storage::DomainDef,
774 target: &str,
775) -> Result<(), EvalError> {
776 let name = target;
777 for chk in &dom.checks {
778 let src = &chk.expr;
779 // v7.39 (round 260) — report the constraint that failed by NAME.
780 let owner = chk.name.as_str();
781 let expr = spg_sql::parser::parse_expression(src).map_err(|e| EvalError::TypeMismatch {
782 detail: alloc::format!("domain {name} CHECK ({src:?}) failed to re-parse: {e:?}"),
783 })?;
784 let synth_cols = alloc::vec![spg_storage::ColumnSchema::new(
785 "value",
786 dom.base_type,
787 dom.nullable,
788 )];
789 let synth_ctx = EvalContext::new(&synth_cols, None);
790 // Owned copy so the temporary row doesn't borrow `v`'s lifetime.
791 let synth_row = spg_storage::Row {
792 values: alloc::vec![v.clone().into_owned()],
793 };
794 let r = eval_expr(&expr, &synth_row, &synth_ctx)?;
795 if matches!(r, Value::Bool(false)) {
796 return Err(EvalError::TypeMismatch {
797 detail: alloc::format!(
798 "value for domain {name} violates check constraint \"{owner}\""
799 ),
800 });
801 }
802 }
803 Ok(())
804}
805
806/// v7.38 (read01 P6.67) — validate a value cast to a user ENUM: a text label
807/// must be one of the enum's members (else error, as PG does); a NULL is a
808/// valid typed null. The stored representation stays the text label.
809/// v7.39 (read01 rowtypes.c) — cast into a user composite type: parse the
810/// `(v1,"v 2",)` record text (double-quote wrapping with doubled quotes,
811/// empty field = NULL) and coerce each field to the declared type; a ROW
812/// value re-labels positionally.
813/// v7.39 (round 350/351, M7 + M11) — how MySQL reads a TEXT operand of
814/// an arithmetic or comparison operator. The identity in the PG dialect,
815/// and out-of-line so it costs the recursive `eval_expr` frame nothing.
816///
817/// Measured on MariaDB 11: `'2024-01-15' + INTERVAL 1 DAY` shifts the
818/// date; `'1abc'+0` is 1, `'abc'+0` is 0, `'2024-01-15'+0` is 2024; two
819/// strings compare as STRINGS (`'10' > '9'` is 0) while a mixed pair
820/// compares numerically (`'10' > 9` is 1).
821#[inline(never)]
822pub(crate) fn mysql_operand_reading_pair(
823 op: BinOp,
824 l: Value<'static>,
825 r: Value<'static>,
826) -> (Value<'static>, Value<'static>) {
827 if !mysql_coerces(op) {
828 return (l, r);
829 }
830 match (&l, &r) {
831 (Value::Text(t), Value::Interval { .. }) => (text_as_temporal(t).unwrap_or(l.clone()), r),
832 (Value::Interval { .. }, Value::Text(t)) => {
833 let rr = text_as_temporal(t).unwrap_or(r.clone());
834 (l, rr)
835 }
836 // v7.39 (round 353, M10) — a boolean IS an integer in MySQL, so
837 // `!1 + 1` is 1 (measured). It was `operator does not exist:
838 // boolean + integer`.
839 (Value::Bool(b), other)
840 if mysql_arith(op) && other.data_type().is_some_and(is_numeric_type) =>
841 {
842 (Value::BigInt(i64::from(*b)), r)
843 }
844 (other, Value::Bool(b))
845 if mysql_arith(op) && other.data_type().is_some_and(is_numeric_type) =>
846 {
847 let rr = Value::BigInt(i64::from(*b));
848 (l, rr)
849 }
850 (Value::Text(t), other) if other.data_type().is_some_and(is_numeric_type) => {
851 (mysql_number_of(t), r)
852 }
853 (other, Value::Text(t)) if other.data_type().is_some_and(is_numeric_type) => {
854 let rr = mysql_number_of(t);
855 (l, rr)
856 }
857 // v7.39 (round 367, M20 P2) — a binary string beside a number
858 // reads as its big-endian integer value (`0x10 + 0` = 16,
859 // `0x10 = 16` is true). Beside a Text operand it stays bytes so
860 // the byte-wise string compare (`0x61 = 'a'`) still fires.
861 (Value::Bytes(b), other) if other.data_type().is_some_and(is_numeric_type) => {
862 (mysql_bytes_as_number(b), r)
863 }
864 (other, Value::Bytes(b)) if other.data_type().is_some_and(is_numeric_type) => {
865 let rr = mysql_bytes_as_number(b);
866 (l, rr)
867 }
868 // Arithmetic between two strings is numeric; comparison is not.
869 (Value::Text(a), Value::Text(b)) if mysql_arith(op) => {
870 (mysql_number_of(a), mysql_number_of(b))
871 }
872 _ => (l, r),
873 }
874}
875
876/// Is this a mixed string/number pair, which MySQL compares numerically?
877fn mysql_mixed_pair(l: &Value<'_>, r: &Value<'_>) -> bool {
878 matches!((l, r), (Value::Text(_), o) | (o, Value::Text(_))
879 if o.data_type().is_some_and(is_numeric_type))
880}
881
882const fn mysql_arith(op: BinOp) -> bool {
883 matches!(
884 op,
885 BinOp::Add | BinOp::Sub | BinOp::Mul | BinOp::Div | BinOp::Mod
886 )
887}
888
889/// Does this comparison need the owned path — because a value's type
890/// wants it, because a CI collation folds it, or (MySQL) because a mixed
891/// string/number pair compares NUMERICALLY there while two strings
892/// compare as strings.
893#[inline(never)]
894fn needs_owned_compare(
895 lc: &Value<'_>,
896 rc: &Value<'_>,
897 lhs: &Expr,
898 rhs: &Expr,
899 ctx: &EvalContext<'_>,
900) -> bool {
901 is_owned_compare_value(lc)
902 || is_owned_compare_value(rc)
903 || compare_is_case_insensitive(lhs, rhs, ctx)
904 || (ctx.mysql_dialect && mysql_mixed_pair(lc, rc))
905}
906
907/// Which operators take MySQL's string→number reading.
908const fn mysql_coerces(op: BinOp) -> bool {
909 matches!(
910 op,
911 BinOp::Add
912 | BinOp::Sub
913 | BinOp::Mul
914 | BinOp::Div
915 | BinOp::Mod
916 | BinOp::Eq
917 | BinOp::NotEq
918 | BinOp::Lt
919 | BinOp::LtEq
920 | BinOp::Gt
921 | BinOp::GtEq
922 )
923}
924
925/// Does this type take part in MySQL's numeric coercion?
926fn is_numeric_type(t: spg_storage::DataType) -> bool {
927 use spg_storage::DataType as D;
928 matches!(
929 t,
930 D::SmallInt | D::Int | D::BigInt | D::Float | D::Real | D::Numeric { .. }
931 )
932}
933
934/// v7.39 (round 364, M4 P2) — a value as it participates in a MySQL
935/// session's default-collation comparison: text folds (accent- and
936/// case-insensitive), everything else is itself. Used by IN and LIKE,
937/// whose comparisons do not pass through `collation_fold_for_compare`.
938fn mysql_collation_key(v: Value<'static>, mysql: bool) -> Value<'static> {
939 match v {
940 Value::Text(s) if mysql => Value::text(spg_storage::mysql_compare_fold(&s)),
941 other => other,
942 }
943}
944
945/// A string as MySQL reads it in numeric position: an exact integer when
946/// the leading number is one, otherwise a double.
947#[inline(never)]
948fn mysql_number_of(s: &str) -> Value<'static> {
949 let n = mysql_leading_number(s);
950 if n.fract() == 0.0 && n.abs() < 9.007_199_254_740_992e15 {
951 #[allow(clippy::cast_possible_truncation)]
952 Value::BigInt(n as i64)
953 } else {
954 Value::Float(n)
955 }
956}
957
958/// v7.39 (round 367, M20 P2) — a MySQL binary string (a `0x…` / `X'…'` /
959/// `b'…'` literal, backed by `Value::Bytes`) reads as its bytes'
960/// BIG-ENDIAN unsigned integer in a numeric context: `0x4142 + 0` is
961/// 16706, `0x10 = 16` is true (measured on MariaDB 11). Only the low 16
962/// bytes participate — a hex literal used in arithmetic is at most an
963/// 8-byte BIGINT in practice — and a value past `i64::MAX` becomes a
964/// NUMERIC so nothing wraps negative.
965fn mysql_bytes_as_number(b: &[u8]) -> Value<'static> {
966 let start = b.len().saturating_sub(16);
967 let acc = b[start..]
968 .iter()
969 .fold(0u128, |a, &x| (a << 8) | u128::from(x));
970 if acc <= i64::MAX as u128 {
971 #[allow(clippy::cast_possible_truncation)]
972 Value::BigInt(acc as i64)
973 } else {
974 crate::conversions::big_literal_to_value(&alloc::format!("{acc}"))
975 }
976}
977
978/// MySQL's `/`: a real division, and NULL on a zero divisor. `None`
979/// when this pairing is not the integer/integer case PG and MySQL
980/// disagree about.
981#[inline(never)]
982pub(crate) fn mysql_true_division(
983 op: BinOp,
984 l: &Value<'_>,
985 r: &Value<'_>,
986 text_operand: bool,
987) -> Option<Value<'static>> {
988 // v7.39 (round 372) — MySQL's `x % 0` / `x MOD 0` is NULL, not the PG
989 // "division by zero" error (measured on MariaDB 11: `10%0`, `10 MOD
990 // 0`, `10.5%0` are all NULL, matching `1/0`). A non-zero divisor takes
991 // the normal modulo path.
992 if matches!(op, BinOp::Mod) {
993 return if value_is_zero(r) {
994 Some(Value::Null)
995 } else {
996 None
997 };
998 }
999 if !matches!(op, BinOp::Div) {
1000 return None;
1001 }
1002 // v7.39 (round 393) — MariaDB `/` on exact (int / decimal) operands is a
1003 // DECIMAL whose scale is the LEFT operand's scale + 4 (`7/2` is 3.5000,
1004 // `10.0/3` is 3.33333, `7.00/2` is 3.500000), NOT a float. A float /
1005 // double operand — or a STRING one, `'10'/'4'` is 2.5 (double) — makes
1006 // the result a float; a zero divisor is NULL.
1007 if text_operand
1008 || matches!(l, Value::Float(_) | Value::Real(_))
1009 || matches!(r, Value::Float(_) | Value::Real(_))
1010 {
1011 let f = |v: &Value<'_>| -> Option<f64> {
1012 match v {
1013 Value::Float(x) => Some(*x),
1014 Value::Real(x) => Some(f64::from(*x)),
1015 Value::SmallInt(n) => Some(f64::from(*n)),
1016 Value::Int(n) => Some(f64::from(*n)),
1017 #[allow(clippy::cast_precision_loss)]
1018 Value::BigInt(n) => Some(*n as f64),
1019 _ => None,
1020 }
1021 };
1022 let (a, b) = (f(l)?, f(r)?);
1023 return Some(if b == 0.0 {
1024 Value::Null
1025 } else {
1026 Value::Float(a / b)
1027 });
1028 }
1029 let (ls, lsc) = exact_decimal_parts(l)?;
1030 let (rs, rsc) = exact_decimal_parts(r)?;
1031 if rs == 0 {
1032 return Some(Value::Null);
1033 }
1034 let result_scale = u32::from(lsc) + 4;
1035 // result_scaled = round( ls * 10^(rsc + 4) / rs ), half away from zero.
1036 let pow = 10i128.checked_pow(u32::from(rsc) + 4)?;
1037 let num = ls.checked_mul(pow)?;
1038 let q = num / rs;
1039 let rem = num % rs;
1040 let bump = if rem.unsigned_abs() * 2 >= rs.unsigned_abs() {
1041 if (num < 0) == (rs < 0) { 1 } else { -1 }
1042 } else {
1043 0
1044 };
1045 Some(Value::numeric(q + bump, u16::try_from(result_scale).ok()?))
1046}
1047
1048/// The `(scaled, scale)` of an exact integer / NUMERIC value: an integer
1049/// has scale 0. None for a float / non-numeric (they take the float path).
1050fn exact_decimal_parts(v: &Value<'_>) -> Option<(i128, u16)> {
1051 match v {
1052 Value::SmallInt(n) => Some((i128::from(*n), 0)),
1053 Value::Int(n) => Some((i128::from(*n), 0)),
1054 Value::BigInt(n) => Some((i128::from(*n), 0)),
1055 Value::Numeric {
1056 scaled,
1057 scale,
1058 kind: spg_storage::NumericKind::Finite,
1059 } => Some((*scaled, *scale)),
1060 _ => None,
1061 }
1062}
1063
1064/// v7.39 (round 383) — the UNSIGNED 64-bit value a MySQL bitwise operand
1065/// reads as. MySQL's `& | ^ ~ << >>` all work on `BIGINT UNSIGNED`, so an
1066/// operand is its 64-bit two's-complement pattern (a negative integer:
1067/// `-5` is `0xFFFF…FB`), rounded to the nearest integer (a float / numeric:
1068/// `2.9` is 3), its big-endian value (a `0x…` binary string), or its
1069/// leading number (a string). Anything else (an inet, a range, a
1070/// bit-string) returns None so the operator keeps its own meaning.
1071#[allow(clippy::cast_possible_truncation, clippy::cast_sign_loss)]
1072fn mysql_bit_u64(v: &Value<'_>) -> Option<u64> {
1073 match v {
1074 Value::SmallInt(n) => Some(i64::from(*n) as u64),
1075 Value::Int(n) => Some(i64::from(*n) as u64),
1076 Value::BigInt(n) => Some(*n as u64),
1077 Value::Bool(b) => Some(u64::from(*b)),
1078 Value::Float(x) => Some(x.round() as i64 as u64),
1079 Value::Real(x) => Some(f64::from(*x).round() as i64 as u64),
1080 Value::Numeric {
1081 scaled,
1082 scale,
1083 kind: spg_storage::NumericKind::Finite,
1084 } => {
1085 // Round half away from zero to an integer, then take the low
1086 // 64 bits (two's complement) — `~2.9` is `~3`.
1087 if *scale > 38 {
1088 return None;
1089 }
1090 let div = 10i128.pow(u32::from(*scale));
1091 let q = scaled / div;
1092 let rem = scaled % div;
1093 let rounded = if rem.unsigned_abs() * 2 >= div.unsigned_abs() {
1094 q + scaled.signum()
1095 } else {
1096 q
1097 };
1098 Some(rounded as u64)
1099 }
1100 Value::Bytes(b) => mysql_bit_u64(&mysql_bytes_as_number(b)),
1101 Value::Text(s) => mysql_bit_u64(&mysql_number_of(s)),
1102 _ => None,
1103 }
1104}
1105
1106/// A MySQL bitwise result — a `BIGINT UNSIGNED`. It stays a signed
1107/// `BigInt` while it fits (so an integer-context consumer — MAKE_SET
1108/// bits, ELT / SUBSTRING / REPEAT counts — still takes it); a value past
1109/// `i64::MAX` (a set bit 63, e.g. `~5`) has no signed integer type, so it
1110/// becomes a scale-0 NUMERIC, which holds the whole `0..=2^64-1` range and
1111/// renders as the plain integer MySQL prints.
1112fn u64_as_value(n: u64) -> Value<'static> {
1113 match i64::try_from(n) {
1114 Ok(v) => Value::BigInt(v),
1115 Err(_) => Value::numeric(i128::from(n), 0),
1116 }
1117}
1118
1119/// v7.39 (round 383) — the MySQL bitwise operators on UNSIGNED 64-bit
1120/// integers. Returns None when either operand is not number-like (so an
1121/// `inet << int` / `bit(n) & bit(n)` / geometric `#` keeps its own path)
1122/// or the operator is not bitwise. A shift of 64 or more is 0 (MySQL does
1123/// not mask the shift count).
1124pub(crate) fn mysql_bitwise(op: BinOp, l: &Value<'_>, r: &Value<'_>) -> Option<Value<'static>> {
1125 let out = match op {
1126 BinOp::BitAnd => mysql_bit_u64(l)? & mysql_bit_u64(r)?,
1127 BinOp::BitOr => mysql_bit_u64(l)? | mysql_bit_u64(r)?,
1128 BinOp::BitXor => mysql_bit_u64(l)? ^ mysql_bit_u64(r)?,
1129 // `<<` / `>>` share the inet-containment BinOps; a numeric pair is a
1130 // shift, anything else stays inet / range / bit-string.
1131 BinOp::InetContainedBy => {
1132 let (a, n) = (mysql_bit_u64(l)?, mysql_bit_u64(r)?);
1133 if n >= 64 { 0 } else { a << n }
1134 }
1135 BinOp::InetContains => {
1136 let (a, n) = (mysql_bit_u64(l)?, mysql_bit_u64(r)?);
1137 if n >= 64 { 0 } else { a >> n }
1138 }
1139 _ => return None,
1140 };
1141 Some(u64_as_value(out))
1142}
1143
1144/// v7.39 (round 383) — MySQL unary `~x`: the UNSIGNED 64-bit complement
1145/// (`~5` is 18446744073709551610). None for a non-number operand (a
1146/// bit-string / inet / macaddr keeps PG's typed complement).
1147pub(crate) fn mysql_bit_not(v: &Value<'_>) -> Option<Value<'static>> {
1148 Some(u64_as_value(!mysql_bit_u64(v)?))
1149}
1150
1151/// v7.39 (round 390, type-fidelity epic P5) — the inline `SET('a','b',…)`
1152/// variant list an expression's column is declared with, or None. Mirrors
1153/// `expr_enum_type_name` — a bare `Expr::Column` looked up by name.
1154pub(crate) fn expr_set_variants<'e>(
1155 e: &'e Expr,
1156 columns: &'e [ColumnSchema],
1157) -> Option<&'e [String]> {
1158 match e {
1159 Expr::Column(c) => columns
1160 .iter()
1161 .find(|col| col.name == c.name)
1162 .and_then(|col| col.inline_set_variants.as_deref()),
1163 _ => None,
1164 }
1165}
1166
1167/// v7.39 (round 402) — the inline `ENUM('a','b',…)` variant list an
1168/// expression's column is declared with, or None. Like `expr_set_variants`.
1169pub(crate) fn expr_inline_enum_variants<'e>(
1170 e: &'e Expr,
1171 columns: &'e [ColumnSchema],
1172) -> Option<&'e [String]> {
1173 match e {
1174 Expr::Column(c) => columns
1175 .iter()
1176 .find(|col| col.name == c.name)
1177 .and_then(|col| col.inline_enum_variants.as_deref()),
1178 _ => None,
1179 }
1180}
1181
1182/// The 1-based ordinal a stored inline-ENUM text carries in a numeric
1183/// context (`e + 0` is 1 for the first member); the empty string / an
1184/// unknown member is 0 (MySQL's implicit `''` enum error value).
1185pub(crate) fn enum_text_to_ordinal(text: &str, variants: &[String]) -> i64 {
1186 variants
1187 .iter()
1188 .position(|v| v == text)
1189 .map_or(0, |p| p as i64 + 1)
1190}
1191
1192/// The bitmask a stored SET text carries in a numeric context: each
1193/// comma-separated member contributes `1 << its position` in the declared
1194/// variant list (`'a,c'` over `('a','b','c','d')` is 1 | 4 = 5). An empty
1195/// string is 0; an unknown member (should not occur — the write path
1196/// validates) contributes nothing.
1197pub(crate) fn set_text_to_bitmask(text: &str, variants: &[String]) -> i64 {
1198 if text.is_empty() {
1199 return 0;
1200 }
1201 let mut bits = 0i64;
1202 for member in text.split(',') {
1203 if let Some(pos) = variants.iter().position(|v| v == member) {
1204 bits |= 1i64 << pos;
1205 }
1206 }
1207 bits
1208}
1209
1210/// Is this an arithmetic / bitwise operator MySQL evaluates a SET column
1211/// numerically under? (`s + 0`, `s & flag`, …). The comparison operators
1212/// are NOT here — `s = 'a,c'` stays a text compare.
1213pub(crate) const fn is_mysql_numeric_binop(op: BinOp) -> bool {
1214 matches!(
1215 op,
1216 BinOp::Add
1217 | BinOp::Sub
1218 | BinOp::Mul
1219 | BinOp::Div
1220 | BinOp::Mod
1221 | BinOp::BitAnd
1222 | BinOp::BitOr
1223 | BinOp::BitXor
1224 | BinOp::InetContainedBy
1225 | BinOp::InetContains
1226 )
1227}
1228
1229/// v7.39 (round 372) — is `v` a numeric zero (any width / kind)? Used to
1230/// route `x % 0` / `MOD(x, 0)` to NULL under the MySQL dialect.
1231pub(crate) fn value_is_zero(v: &Value<'_>) -> bool {
1232 match v {
1233 Value::SmallInt(n) => *n == 0,
1234 Value::Int(n) => *n == 0,
1235 Value::BigInt(n) => *n == 0,
1236 Value::Float(x) => *x == 0.0,
1237 Value::Real(x) => *x == 0.0,
1238 Value::Numeric { scaled, .. } => *scaled == 0,
1239 _ => false,
1240 }
1241}
1242
1243/// `-'5'` in the MySQL dialect. Out-of-line for the same frame reason.
1244#[inline(never)]
1245fn mysql_negate_text(
1246 op: spg_sql::ast::UnOp,
1247 v: &Value<'static>,
1248) -> Option<Result<Value<'static>, EvalError>> {
1249 match v {
1250 Value::Text(t) => Some(apply_unary(op, mysql_number_of(t))),
1251 _ => None,
1252 }
1253}
1254
1255/// The MySQL reading of a unary operator, or None to let `apply_unary` (the
1256/// PG path) run. Kept out of `eval_expr`'s recursive frame — see the
1257/// round-383 frame cliff. Covers `NOT` on any truth value (round 346),
1258/// `-'str'` numeric negation (round 351), and the unsigned `~` complement
1259/// (round 383).
1260#[inline(never)]
1261fn mysql_unary_arm(
1262 op: spg_sql::ast::UnOp,
1263 v: &Value<'static>,
1264) -> Option<Result<Value<'static>, EvalError>> {
1265 use spg_sql::ast::UnOp;
1266 match op {
1267 // `NOT 5` is 0 — read any non-bool as a truth value; a bool / NULL
1268 // keeps the PG path (still refused there for non-bool).
1269 UnOp::Not if !matches!(v, Value::Bool(_) | Value::Null) => Some(mysql_not(v)),
1270 // `-'5'` is -5, `-'abc'` is 0.
1271 UnOp::Neg => mysql_negate_text(op, v),
1272 // `+ anything` is that thing: measured on MariaDB 11, `+'x'` is
1273 // 'x', `+TRUE` is 1, `+NULL` is NULL. No type check at all, unlike
1274 // PG, which refuses every non-numeric operand.
1275 UnOp::Plus => Some(Ok(v.clone())),
1276 // `~5` is the unsigned 64-bit complement; NULL stays NULL (PG path).
1277 UnOp::BitNot if !matches!(v, Value::Null) => mysql_bit_not(v).map(Ok),
1278 _ => None,
1279 }
1280}
1281
1282/// A date / timestamp string as its temporal value, or `None` when it is
1283/// not one (in which case the operand is left exactly as it was).
1284#[inline(never)]
1285fn text_as_temporal(t: &str) -> Option<Value<'static>> {
1286 parse_timestamp_literal(t)
1287 .map(Value::Timestamp)
1288 .or_else(|| parse_date_literal(t).map(Value::Date))
1289}
1290
1291/// v7.39 (round 620) — an unadorned string literal, which is what PG calls
1292/// `unknown`: a value whose type the context gets to choose. `''::TEXT` is
1293/// not one, and neither is a text column.
1294fn is_unknown_string_literal(e: &Expr) -> bool {
1295 matches!(e, Expr::Literal(spg_sql::ast::Literal::String(_)))
1296}
1297
1298/// v7.39 (round 620) — resolve such a literal to boolean, which is what a
1299/// boolean connective asks of it. An unparseable one is PG's input-syntax
1300/// error (22P02), not a type complaint: `'a' AND true` says
1301/// `invalid input syntax for type boolean: "a"`, exactly as `'a'::BOOLEAN`
1302/// does — same failure, same words, because it is the same coercion.
1303#[inline(never)]
1304fn coerce_unknown_literal_to_bool(e: &Expr) -> Result<Value<'static>, EvalError> {
1305 let Expr::Literal(spg_sql::ast::Literal::String(s)) = e else {
1306 unreachable!("guarded by is_unknown_string_literal")
1307 };
1308 cast::cast_value_in(
1309 Value::Text(s.clone().into()),
1310 spg_sql::ast::CastTarget::Bool,
1311 false,
1312 )
1313}
1314
1315/// v7.39 (round 621) — a literal that is plainly not a boolean, and the PG
1316/// type name for it. `NULL` and a bare string literal are deliberately absent:
1317/// neither carries a type of its own, and a boolean connective is a context
1318/// that gives them one.
1319fn non_boolean_literal_type(e: &Expr) -> Option<&'static str> {
1320 use spg_sql::ast::Literal as L;
1321 match e {
1322 Expr::Literal(L::Integer(_)) => Some("integer"),
1323 Expr::Literal(L::Float(_)) => Some("double precision"),
1324 Expr::Literal(L::Numeric { .. } | L::NumericBig(_)) => Some("numeric"),
1325 _ => None,
1326 }
1327}
1328
1329/// v7.39 (round 621) — `AND` / `OR`, evaluated the way PG evaluates them.
1330///
1331/// Round 620 handled the unknown literal here; round 621 adds the part that
1332/// makes `WHERE x <> 0 AND 1/x > 0` work at all. SPG evaluated both sides
1333/// always, so the guard idiom — the whole reason that predicate is written
1334/// that way — raised on the very rows the guard exists to exclude. Measured
1335/// against PG: `false AND (1/0 = 0)` answers `f`, `true OR (1/0 = 0)` answers
1336/// `t`, and a filter guarded that way returns its rows.
1337///
1338/// PG affords that AND still refuses `false AND 1`, because the two happen at
1339/// different times: the operand types are checked during ANALYSIS, before any
1340/// evaluation, and the short circuit is a RUN-TIME decision. Both parts are
1341/// here — the right-hand operand's type is read statically (it is the side
1342/// that may go unevaluated), and only a type that is definitively known and
1343/// definitively not boolean is refused. An unknown type is left alone, so a
1344/// shape the describer cannot type keeps the old behaviour rather than
1345/// earning a spurious error.
1346///
1347/// Order is PG's too, and it is strictly left-first: `(1/0 = 0) AND false`
1348/// raises on both, because the left is evaluated before anything can decide
1349/// that it did not need to be.
1350///
1351/// Out-of-line so it costs `eval_expr` no frame (the round-305 frame cliff).
1352#[inline(never)]
1353fn eval_connective(
1354 lhs: &Expr,
1355 op: BinOp,
1356 rhs: &Expr,
1357 row: &Row<'static>,
1358 ctx: &EvalContext<'_>,
1359) -> Result<Value<'static>, EvalError> {
1360 let side = |e: &Expr| -> Result<Value<'static>, EvalError> {
1361 if is_unknown_string_literal(e) {
1362 coerce_unknown_literal_to_bool(e)
1363 } else {
1364 eval_expr(e, row, ctx)
1365 }
1366 };
1367 let l = side(lhs)?;
1368 // The analysis-time half: refuse a right-hand operand that is plainly not
1369 // boolean, whether or not the short circuit would reach it.
1370 //
1371 // Only a LITERAL is read this way. The first cut asked
1372 // `describe_expr_type` for any expression's type, and it answers
1373 // confidently and wrongly for shapes that matter here — `NULL` comes back
1374 // as text, so `true AND NULL` earned a type error; and a MATCH … AGAINST
1375 // folds internally into an OR over tsvector operands, so full-text search
1376 // stopped working. Three existing pins caught all of it. A literal cannot
1377 // be misread, and it is what PG's own refusals in this area are about.
1378 if let Some(ty) = non_boolean_literal_type(rhs) {
1379 return Err(EvalError::TypeMismatch {
1380 detail: alloc::format!(
1381 "argument of {} must be type boolean, not type {ty}",
1382 if matches!(op, BinOp::And) {
1383 "AND"
1384 } else {
1385 "OR"
1386 },
1387 ),
1388 });
1389 }
1390 // Resolving an unknown literal belongs to the same half — it is a
1391 // coercion PG performs while analysing, so `false AND 'a'` says
1392 // `invalid input syntax for type boolean: "a"` rather than answering `f`.
1393 let rhs_resolved = if is_unknown_string_literal(rhs) {
1394 Some(coerce_unknown_literal_to_bool(rhs)?)
1395 } else {
1396 None
1397 };
1398 // The run-time half.
1399 match (op, &l) {
1400 (BinOp::And, Value::Bool(false)) => return Ok(Value::Bool(false)),
1401 (BinOp::Or, Value::Bool(true)) => return Ok(Value::Bool(true)),
1402 _ => {}
1403 }
1404 let r = match rhs_resolved {
1405 Some(v) => v,
1406 None => side(rhs)?,
1407 };
1408 if matches!(op, BinOp::And) {
1409 and_3vl(l, r)
1410 } else {
1411 apply_binary(op, l, r)
1412 }
1413}
1414
1415/// v7.39 (round 346, M1) — the MySQL reading of `AND` / `OR`, out-of-line
1416/// so it costs `eval_expr` no frame (see the round-305 frame cliff).
1417#[inline(never)]
1418fn eval_mysql_connective(
1419 lhs: &Expr,
1420 op: BinOp,
1421 rhs: &Expr,
1422 row: &Row<'static>,
1423 ctx: &EvalContext<'_>,
1424) -> Result<Value<'static>, EvalError> {
1425 let l = as_mysql_truth(eval_expr(lhs, row, ctx)?)?;
1426 let r = as_mysql_truth(eval_expr(rhs, row, ctx)?)?;
1427 apply_mysql_connective(op, l, r)
1428}
1429
1430/// v7.39 (round 407) — apply a MySQL logical connective (`AND` / `OR` /
1431/// `XOR`) to two operands already reduced to truth values (`Bool` or
1432/// `Null`). AND / OR reuse the dialect-blind `apply_binary`; `XOR` is
1433/// MySQL-only (no `apply_binary` arm) and computed here: NULL on either
1434/// side yields NULL, otherwise the exclusive-or of the two truth values.
1435pub(crate) fn apply_mysql_connective(
1436 op: BinOp,
1437 l: Value<'static>,
1438 r: Value<'static>,
1439) -> Result<Value<'static>, EvalError> {
1440 if op == BinOp::LogicalXor {
1441 return Ok(match (&l, &r) {
1442 (Value::Bool(a), Value::Bool(b)) => Value::Bool(a != b),
1443 _ => Value::Null,
1444 });
1445 }
1446 apply_binary(op, l, r)
1447}
1448
1449#[inline(never)]
1450pub(crate) fn as_mysql_truth(v: Value<'static>) -> Result<Value<'static>, EvalError> {
1451 Ok(match v {
1452 Value::Null => Value::Null,
1453 other => Value::Bool(predicate_is_true(&other, "AND", true)?),
1454 })
1455}
1456
1457/// The MySQL reading of `NOT`, likewise out-of-line.
1458#[inline(never)]
1459fn mysql_not(v: &Value<'_>) -> Result<Value<'static>, EvalError> {
1460 Ok(Value::Bool(!predicate_is_true(v, "NOT", true)?))
1461}
1462
1463/// v7.39 (round 346, M1) — is this value TRUE, in a position that wants a
1464/// truth value (WHERE / CASE WHEN / NOT / AND / OR / HAVING / ON)?
1465///
1466/// The engine used to write `matches!(v, Value::Bool(true))` at every such
1467/// position, so anything that was not already a boolean silently read as
1468/// FALSE. `SELECT CASE WHEN 1 THEN 'a' END` answered NULL and — far worse —
1469/// `SELECT … WHERE 1` returned **no rows at all**. Neither dialect does
1470/// that: MariaDB 11 takes any non-zero number as true, and PG 18.4 raises
1471/// `argument of WHERE must be type boolean, not type integer`.
1472///
1473/// NULL is not true (three-valued logic) and is not an error in either.
1474pub(crate) fn predicate_is_true(v: &Value<'_>, kw: &str, mysql: bool) -> Result<bool, EvalError> {
1475 match v {
1476 Value::Bool(b) => Ok(*b),
1477 Value::Null => Ok(false),
1478 _ if mysql => Ok(mysql_truthy(v)),
1479 // PG resolves a bare literal in this position through boolean
1480 // INPUT, so `CASE WHEN 'true'` is legal and `'abc'` is not.
1481 Value::Text(t) => match crate::eval::cast::cast_value(
1482 Value::text(t.to_string()),
1483 spg_sql::ast::CastTarget::Bool,
1484 )? {
1485 Value::Bool(b) => Ok(b),
1486 _ => Ok(false),
1487 },
1488 other => Err(EvalError::TypeMismatch {
1489 detail: alloc::format!(
1490 "argument of {kw} must be type boolean, not type {}",
1491 crate::eval::strings::pg_typeof_name(other)
1492 ),
1493 }),
1494 }
1495}
1496
1497/// MariaDB 11's reading, measured: a number is true when it is not zero
1498/// (`-1` and `0.5` are both true); a string contributes its LEADING
1499/// number, so `'1abc'` is true while `'abc'` and `''` are false.
1500fn mysql_truthy(v: &Value<'_>) -> bool {
1501 match v {
1502 Value::Bool(b) => *b,
1503 Value::Null => false,
1504 Value::SmallInt(n) => *n != 0,
1505 Value::Int(n) => *n != 0,
1506 Value::BigInt(n) => *n != 0,
1507 Value::Float(f) => *f != 0.0,
1508 Value::Real(f) => *f != 0.0,
1509 Value::Numeric { scaled, .. } => *scaled != 0,
1510 Value::Text(t) => mysql_leading_number(t) != 0.0,
1511 Value::BpChar(t) => mysql_leading_number(t) != 0.0,
1512 // Everything else converts to a non-zero number in MariaDB (a
1513 // DATE reads as its YYYYMMDD digits, for one).
1514 _ => true,
1515 }
1516}
1517
1518/// The leading numeric prefix of a string, MySQL-style: `'1abc'` is 1,
1519/// `'abc'` and `''` are 0.
1520#[inline(never)]
1521pub(crate) fn mysql_leading_number(s: &str) -> f64 {
1522 let t = s.trim_start();
1523 let mut end = 0usize;
1524 let mut seen_dot = false;
1525 let mut seen_digit = false;
1526 // v7.39 (round 351, M11) — the exponent form counts: MariaDB reads
1527 // `'1e3'` as 1000 and `'1.5e2'` as 150 (measured). A trailing `e`
1528 // with no digits after it is not part of the number (`'1e'` is 1).
1529 let mut seen_exp = false;
1530 let mut exp_at = 0usize;
1531 for (i, c) in t.char_indices() {
1532 match c {
1533 '-' | '+' if i == 0 => {}
1534 '-' | '+' if seen_exp && i == exp_at + 1 => {}
1535 '0'..='9' => seen_digit = true,
1536 '.' if !seen_dot && !seen_exp => seen_dot = true,
1537 'e' | 'E' if seen_digit && !seen_exp => {
1538 seen_exp = true;
1539 exp_at = i;
1540 }
1541 _ => break,
1542 }
1543 end = i + c.len_utf8();
1544 }
1545 if !seen_digit {
1546 return 0.0;
1547 }
1548 // Trim an exponent that never got its digits.
1549 let mut text = &t[..end];
1550 while !text.is_empty() && text.parse::<f64>().is_err() {
1551 text = &text[..text.len() - 1];
1552 }
1553 text.parse::<f64>().unwrap_or(0.0)
1554}
1555
1556/// v7.39 (read01 ruleutils.c) — resolve a relation name to its synthetic
1557/// oid: user tables in the 16384+ band (table_names order), views at
1558/// 32768+, and the synthesised system catalogs at their REAL PG oids.
1559/// `None` when the name is unknown (the caller keeps the legacy text
1560/// behaviour so `'anything'::regclass::text` still round-trips).
1561pub(crate) fn regclass_name_to_oid(cat: &spg_storage::Catalog, bare: &str) -> Option<i64> {
1562 // v7.39 (round 337, V62) — an INDEX and a SEQUENCE are relations too:
1563 // both have a `pg_class` row, so both answer to `::regclass` in PG.
1564 // v7.39 (round 338, V64) — and the bands live in ONE allocator now,
1565 // shared with the catalog synths, so `pg_class.oid = 'x'::regclass`
1566 // holds for every kind rather than only for tables.
1567 if let Some(oid) = crate::system_catalog::relation_oid(cat, bare) {
1568 return Some(oid);
1569 }
1570 Some(match bare {
1571 "pg_type" => 1247,
1572 "pg_attribute" => 1249,
1573 "pg_proc" => 1255,
1574 "pg_class" => 1259,
1575 "pg_database" => 1262,
1576 "pg_constraint" => 2606,
1577 "pg_index" => 2610,
1578 "pg_namespace" => 2615,
1579 // v7.39 (round 650) — the text-search catalogs. This list is a
1580 // hand-kept subset of `CATALOG_RELATIONS`, which is why adding a
1581 // catalog there was not enough for `'pg_ts_config'::regclass`.
1582 "pg_ts_config" => 3602,
1583 "pg_ts_config_map" => 3603,
1584 "pg_ts_dict" => 3600,
1585 "pg_ts_parser" => 3601,
1586 "pg_ts_template" => 3764,
1587 _ => return None,
1588 })
1589}
1590
1591/// v7.39 (round 263) — crate-visible wrapper so the write path can
1592/// relabel + coerce a value into a composite column's declared type.
1593pub(crate) fn apply_composite_cast_pub(
1594 v: Value<'static>,
1595 comp: &spg_storage::CompositeDef,
1596 cat: Option<&spg_storage::Catalog>,
1597) -> Result<Value<'static>, EvalError> {
1598 apply_composite_cast_in(v, comp, cat)
1599}
1600
1601/// v7.39 (round 264) — resolve one field's value, recursing when the
1602/// field is itself a COMPOSITE. Without this a nested field kept the
1603/// inner record's TEXT rendering, so `(x).inner.street` errored and
1604/// `row_to_json` nested a string rather than an object.
1605fn coerce_composite_field(
1606 val: Value<'static>,
1607 fname: &str,
1608 fty: spg_storage::DataType,
1609 user_ty: Option<&str>,
1610 cat: Option<&spg_storage::Catalog>,
1611) -> Result<Value<'static>, EvalError> {
1612 if matches!(val, Value::Null) {
1613 return Ok(val);
1614 }
1615 if let Some(tn) = user_ty
1616 && let Some(inner) = cat.and_then(|c| c.composite_types().get(tn))
1617 {
1618 return apply_composite_cast_in(val, inner, cat);
1619 }
1620 crate::conversions::coerce_value(val, fty, fname, 0).map_err(|e| EvalError::TypeMismatch {
1621 detail: alloc::format!("{e}"),
1622 })
1623}
1624
1625fn apply_composite_cast(
1626 v: Value<'static>,
1627 comp: &spg_storage::CompositeDef,
1628) -> Result<Value<'static>, EvalError> {
1629 apply_composite_cast_in(v, comp, None)
1630}
1631
1632fn apply_composite_cast_in(
1633 v: Value<'static>,
1634 comp: &spg_storage::CompositeDef,
1635 cat: Option<&spg_storage::Catalog>,
1636) -> Result<Value<'static>, EvalError> {
1637 match v {
1638 Value::Null => Ok(Value::Null),
1639 Value::Composite(fields) => {
1640 if fields.len() != comp.fields.len() {
1641 // PG reports the SHAPE mismatch as a plain cast refusal.
1642 return Err(EvalError::TypeMismatch {
1643 detail: alloc::format!("cannot cast type record to {}", comp.name),
1644 });
1645 }
1646 // v7.39 (round 263) — relabel AND coerce: this branch only
1647 // renamed the fields, so `ROW('x','notanint')::addr` kept the
1648 // text in an int field and PG's input error never fired.
1649 let mut out: alloc::vec::Vec<(alloc::string::String, Value<'static>)> =
1650 alloc::vec::Vec::with_capacity(comp.fields.len());
1651 for (i, ((name, fty), (_, val))) in comp.fields.iter().zip(fields).enumerate() {
1652 let ut = comp.field_user_types.get(i).and_then(Option::as_deref);
1653 let coerced = coerce_composite_field(val, name, *fty, ut, cat)?;
1654 out.push((name.clone(), coerced));
1655 }
1656 Ok(Value::Composite(out))
1657 }
1658 Value::Text(s) => {
1659 let raw = parse_record_text(s.as_ref()).ok_or_else(|| EvalError::TypeMismatch {
1660 detail: alloc::format!("malformed record literal: \"{s}\""),
1661 })?;
1662 if raw.len() != comp.fields.len() {
1663 return Err(EvalError::TypeMismatch {
1664 detail: alloc::format!("malformed record literal: \"{s}\""),
1665 });
1666 }
1667 let mut out: alloc::vec::Vec<(alloc::string::String, Value<'static>)> =
1668 alloc::vec::Vec::with_capacity(raw.len());
1669 for (i, ((fname, fty), field_text)) in comp.fields.iter().zip(raw).enumerate() {
1670 let ut = comp.field_user_types.get(i).and_then(Option::as_deref);
1671 let val = match field_text {
1672 None => Value::Null,
1673 Some(t) => coerce_composite_field(Value::text(t), fname, *fty, ut, cat)?,
1674 };
1675 out.push((fname.clone(), val));
1676 }
1677 Ok(Value::Composite(out))
1678 }
1679 other => Err(EvalError::TypeMismatch {
1680 detail: alloc::format!(
1681 "cannot cast {} to composite type \"{}\"",
1682 crate::conversions::pg_type_name_for_error_opt(other.data_type()),
1683 comp.name
1684 ),
1685 }),
1686 }
1687}
1688
1689/// Split PG's record text `(f1,f2,...)` into per-field raw strings
1690/// (None = empty field = NULL). Double quotes wrap fields containing
1691/// metacharacters; `""` inside is a literal quote; a backslash escapes
1692/// the next character.
1693fn parse_record_text(s: &str) -> Option<alloc::vec::Vec<Option<alloc::string::String>>> {
1694 let t = s.trim();
1695 let inner = t.strip_prefix('(')?.strip_suffix(')')?;
1696 let mut out: alloc::vec::Vec<Option<alloc::string::String>> = alloc::vec::Vec::new();
1697 let chars: alloc::vec::Vec<char> = inner.chars().collect();
1698 let mut field = alloc::string::String::new();
1699 let mut quoted_seen = false;
1700 let mut i = 0usize;
1701 let mut in_quotes = false;
1702 loop {
1703 if i >= chars.len() {
1704 if in_quotes {
1705 return None;
1706 }
1707 out.push(if field.is_empty() && !quoted_seen {
1708 None
1709 } else {
1710 Some(field.clone())
1711 });
1712 break;
1713 }
1714 let c = chars[i];
1715 if in_quotes {
1716 match c {
1717 '"' if chars.get(i + 1) == Some(&'"') => {
1718 field.push('"');
1719 i += 2;
1720 }
1721 '"' => {
1722 in_quotes = false;
1723 i += 1;
1724 }
1725 '\\' => {
1726 field.push(*chars.get(i + 1)?);
1727 i += 2;
1728 }
1729 _ => {
1730 field.push(c);
1731 i += 1;
1732 }
1733 }
1734 } else {
1735 match c {
1736 '"' => {
1737 in_quotes = true;
1738 quoted_seen = true;
1739 i += 1;
1740 }
1741 ',' => {
1742 out.push(if field.is_empty() && !quoted_seen {
1743 None
1744 } else {
1745 Some(core::mem::take(&mut field))
1746 });
1747 quoted_seen = false;
1748 i += 1;
1749 }
1750 '\\' => {
1751 field.push(*chars.get(i + 1)?);
1752 i += 2;
1753 }
1754 _ => {
1755 field.push(c);
1756 i += 1;
1757 }
1758 }
1759 }
1760 }
1761 Some(out)
1762}
1763
1764fn apply_enum_cast<'a>(
1765 v: Value<'a>,
1766 en: &spg_storage::EnumDef,
1767 name: &str,
1768) -> Result<Value<'a>, EvalError> {
1769 match &v {
1770 Value::Null => Ok(v),
1771 Value::Text(s) => {
1772 if en.labels.iter().any(|l| l.as_str() == s.as_ref()) {
1773 Ok(v)
1774 } else {
1775 Err(EvalError::TypeMismatch {
1776 detail: alloc::format!("invalid input value for enum {name}: {s:?}"),
1777 })
1778 }
1779 }
1780 other => Err(EvalError::TypeMismatch {
1781 detail: alloc::format!(
1782 "cannot cast {} to enum {name}",
1783 crate::conversions::pg_type_name_for_error_opt(other.data_type())
1784 ),
1785 }),
1786 }
1787}
1788
1789/// v7.39 (read01 utils/adt, enum.c) — enum_first / enum_last /
1790/// enum_range resolved from the argument's STATIC enum type (an explicit
1791/// `::enumtype` cast or a column's `ColumnSchema.user_enum_type`) over the
1792/// catalog's member order. Returns None when no argument names a known
1793/// enum, letting the generic function path produce its usual error.
1794/// Out-of-line (`inline(never)`) so the sizable locals don't land in
1795/// `eval_expr`'s recursion frame.
1796/// Enum-ness lives outside the DataType lattice: the witness for "this
1797/// expression is enum-typed" is an explicit `::enumtype` cast or a column
1798/// whose `ColumnSchema.user_enum_type` is set.
1799/// v7.39 (round 258) — crate-visible wrapper so the projection builder
1800/// can keep an expression's enum identity (see `select.rs`).
1801pub(crate) fn expr_enum_type_name_pub<'e>(
1802 e: &'e Expr,
1803 columns: &'e [ColumnSchema],
1804) -> Option<&'e str> {
1805 expr_enum_type_name(e, columns)
1806}
1807
1808/// v7.39 (round 425) — the fractional-seconds precision a projected
1809/// expression should RENDER with: the widest declared precision among the
1810/// MySQL temporal columns it reads. `MAX(d3)` and `d3 + INTERVAL 1 SECOND`
1811/// both keep `d3`'s three digits, as MariaDB does. `None` when the
1812/// expression touches no such column, which leaves PG rendering untouched.
1813///
1814/// Residual (recorded, not modelled): MariaDB also WIDENS the precision from
1815/// some operands — `DATE_ADD(d3, INTERVAL 1 MICROSECOND)` prints six digits
1816/// there. Taking the max over referenced columns covers the common shapes
1817/// and never narrows below the source column.
1818pub(crate) fn expr_mysql_fsp(e: &Expr, columns: &[ColumnSchema]) -> Option<u8> {
1819 fn walk(e: &Expr, columns: &[ColumnSchema], best: &mut Option<u8>) {
1820 match e {
1821 Expr::Column(c) => {
1822 if let Some(f) = columns
1823 .iter()
1824 .find(|col| col.name == c.name)
1825 .and_then(|col| col.mysql_fsp)
1826 {
1827 *best = Some(best.map_or(f, |b: u8| b.max(f)));
1828 }
1829 }
1830 Expr::Binary { lhs, rhs, .. } => {
1831 walk(lhs, columns, best);
1832 walk(rhs, columns, best);
1833 }
1834 Expr::Unary { expr, .. } | Expr::Cast { expr, .. } => walk(expr, columns, best),
1835 Expr::FunctionCall { args, .. } => {
1836 for a in args {
1837 walk(a, columns, best);
1838 }
1839 }
1840 Expr::Case {
1841 operand,
1842 branches,
1843 else_branch,
1844 } => {
1845 if let Some(o) = operand.as_deref() {
1846 walk(o, columns, best);
1847 }
1848 for (w, t) in branches {
1849 walk(w, columns, best);
1850 walk(t, columns, best);
1851 }
1852 if let Some(el) = else_branch.as_deref() {
1853 walk(el, columns, best);
1854 }
1855 }
1856 _ => {}
1857 }
1858 }
1859 let mut best = None;
1860 walk(e, columns, &mut best);
1861 best
1862}
1863
1864/// v7.39 (round 467) — is this expression MySQL-UNSIGNED?
1865///
1866/// MySQL decides unsignedness statically, from the expression's type, not
1867/// from the value it happens to produce. Measured on MariaDB 11: `SUM(a) -
1868/// 100` answers -99 even though `a` is `INT UNSIGNED`, because SUM's result
1869/// type is not unsigned; `a - 5` on the same column raises 1690. So this
1870/// walks the expression the way MySQL's type resolution does.
1871///
1872/// A cast names its target `unsigned` (the parser lowercases MySQL's
1873/// `CAST(x AS UNSIGNED)` into `CastTarget::Named`). Arithmetic is unsigned
1874/// when EITHER operand is — that is MySQL's rule, and it is why `1 - b`
1875/// raises while `5 - a` does not: both are unsigned expressions, but only
1876/// the first has a negative result.
1877///
1878/// Deliberately NOT unsigned: unary minus (MariaDB answers -1 for
1879/// `-CAST(1 AS UNSIGNED)`), and every function result including the
1880/// aggregates. Both measured.
1881pub(crate) fn expr_is_mysql_unsigned(e: &Expr, columns: &[ColumnSchema]) -> bool {
1882 match e {
1883 Expr::Column(c) => columns
1884 .iter()
1885 .find(|col| col.name == c.name)
1886 .is_some_and(|col| col.is_unsigned),
1887 Expr::Cast {
1888 target: CastTarget::Named(n),
1889 ..
1890 } => n.eq_ignore_ascii_case("unsigned"),
1891 Expr::Binary {
1892 lhs,
1893 op: BinOp::Add | BinOp::Sub | BinOp::Mul,
1894 rhs,
1895 } => expr_is_mysql_unsigned(lhs, columns) || expr_is_mysql_unsigned(rhs, columns),
1896 _ => false,
1897 }
1898}
1899
1900/// v7.39 (round 467) — MySQL arithmetic over an UNSIGNED operand, with
1901/// MySQL's range check.
1902///
1903/// `INT UNSIGNED` columns holding 1 and 5 made `a - b` answer **-4** in a
1904/// MySQL session. MariaDB raises `ERROR 1690 (22003): BIGINT UNSIGNED value
1905/// is out of range`. A negative answer where the server promises a
1906/// non-negative one is the kind of thing an application stores back into
1907/// the same column, so it was silent and wrong in the worst direction.
1908///
1909/// The check runs in i128 so the subtraction that underflows is observed
1910/// rather than wrapped, and it only fires when the expression is unsigned
1911/// AND both operands are integers — a NUMERIC or float operand takes the
1912/// ordinary path, as it does in MySQL.
1913///
1914/// `#[inline(never)]`: this is called from the recursive evaluator's
1915/// hottest frame, which already sits against the stack guard.
1916#[inline(never)]
1917fn apply_binary_mysql_unsigned(
1918 op: BinOp,
1919 lhs: &Expr,
1920 rhs: &Expr,
1921 l: Value<'static>,
1922 r: Value<'static>,
1923 ctx: &EvalContext,
1924) -> Result<Value<'static>, EvalError> {
1925 if ctx.mysql_dialect
1926 && matches!(op, BinOp::Add | BinOp::Sub | BinOp::Mul)
1927 && let Some(a) = mysql_int_operand(&l)
1928 && let Some(b) = mysql_int_operand(&r)
1929 && (expr_is_mysql_unsigned(lhs, ctx.columns) || expr_is_mysql_unsigned(rhs, ctx.columns))
1930 {
1931 let out = match op {
1932 BinOp::Add => a.checked_add(b),
1933 BinOp::Sub => a.checked_sub(b),
1934 _ => a.checked_mul(b),
1935 };
1936 let in_range = out.is_some_and(|v| (0..=i128::from(u64::MAX)).contains(&v));
1937 if !in_range {
1938 // MariaDB names the offending expression in the message, with
1939 // minimal parentheses — `a * 0 - 1`, not `((a * 0) - 1)`.
1940 // `pretty_expr` is the deparser that already produces that
1941 // shape. Residual, recorded rather than faked: MariaDB writes
1942 // its columns fully qualified in backticks
1943 // (`db`.`tbl`.`col`), and the database name is not something
1944 // the evaluation context carries.
1945 return Err(EvalError::TypeMismatch {
1946 detail: alloc::format!(
1947 "BIGINT UNSIGNED value is out of range in '{}'",
1948 spg_sql::ast::pretty_expr_mysql(&Expr::Binary {
1949 lhs: alloc::boxed::Box::new(lhs.clone()),
1950 op,
1951 rhs: alloc::boxed::Box::new(rhs.clone()),
1952 })
1953 ),
1954 });
1955 }
1956 }
1957 let out = apply_binary_in(op, l, r, ctx.mysql_dialect);
1958 // v7.39 (round 503) — MariaDB answers NULL for division / modulo by
1959 // zero; SPG raised.
1960 //
1961 // Measured against MariaDB 11: `SELECT 1/0`, `SELECT 5 DIV 0` and
1962 // `SELECT 5 % 0` are all NULL — and they are NULL under the DEFAULT
1963 // sql_mode too, which contains `ERROR_FOR_DIVISION_BY_ZERO`. That flag
1964 // governs WRITES, not the expression: the division evaluates to NULL,
1965 // and a strict-mode INSERT of that result is what raises 1365.
1966 //
1967 // The rule is therefore the DIALECT's, not the mode's: in a MySQL
1968 // session the expression is NULL. It is deliberately not gated on
1969 // `mysql_strict` — an earlier cut of this was, and the gate fired only
1970 // because the probe's context happens to carry no engine, which a
1971 // later round attaching one would have silently reversed.
1972 //
1973 // RESIDUAL, recorded rather than faked: MariaDB's strict-mode INSERT
1974 // of a division by zero raises 1365 and its non-strict INSERT stores
1975 // NULL. SPG's INSERT path evaluates elsewhere and still raises, so it
1976 // matches strict and diverges from non-strict. Closing that needs the
1977 // expression to know it is in a write, which nothing here carries.
1978 if ctx.mysql_dialect && matches!(out, Err(EvalError::DivisionByZero)) {
1979 return Ok(Value::Null);
1980 }
1981 out
1982}
1983
1984/// The integer an operand contributes to the unsigned range check, or
1985/// `None` when it is not an integer at all (NULL, text, NUMERIC, float).
1986fn mysql_int_operand(v: &Value<'_>) -> Option<i128> {
1987 match v {
1988 Value::SmallInt(n) => Some(i128::from(*n)),
1989 Value::Int(n) => Some(i128::from(*n)),
1990 Value::BigInt(n) => Some(i128::from(*n)),
1991 // v7.39 (round 471) — a BIGINT UNSIGNED cell is stored as Numeric
1992 // with scale 0, so the range check has to see it as the integer it
1993 // is. Without this arm the column's own type moved it out of reach
1994 // of round 467's guard and `c - 5` went back to answering -4.
1995 Value::Numeric { scaled, scale, .. } if *scale == 0 => Some(*scaled),
1996 _ => None,
1997 }
1998}
1999
2000fn expr_enum_type_name<'e>(e: &'e Expr, columns: &'e [ColumnSchema]) -> Option<&'e str> {
2001 match e {
2002 Expr::Cast {
2003 target: CastTarget::Named(n),
2004 ..
2005 } => Some(n.as_str()),
2006 Expr::Column(c) => columns
2007 .iter()
2008 .find(|col| col.name == c.name)
2009 // v7.39 (round 259) — a DOMAIN column carries its name in its
2010 // own field; both are "the user type this column is declared
2011 // as", which is what the callers (enum-order comparison,
2012 // pg_typeof) want. Callers gate on the catalog, so a name that
2013 // is one kind never resolves as the other.
2014 .and_then(|col| {
2015 col.user_enum_type
2016 .as_deref()
2017 .or(col.user_domain_type.as_deref())
2018 }),
2019 _ => None,
2020 }
2021}
2022
2023/// v7.39 (enum order knife) — the member-label list for an enum-typed
2024/// expression, or None when the expression carries no enum witness or the
2025/// name is not a known enum. The returned slice borrows the catalog.
2026pub(crate) fn expr_enum_labels<'c>(
2027 e: &Expr,
2028 columns: &[ColumnSchema],
2029 catalog: Option<&'c spg_storage::Catalog>,
2030) -> Option<&'c [String]> {
2031 let name = expr_enum_type_name(e, columns)?;
2032 catalog
2033 .and_then(|cat| cat.enum_types().get(name))
2034 .map(|en| en.labels.as_slice())
2035}
2036
2037/// v7.39 (enum order knife) — compare two enum labels by member order.
2038/// None when either side is not Text or not a member (caller falls back to
2039/// the generic comparison, so a stray value never panics or misorders
2040/// silently differently from before).
2041pub(crate) fn enum_ord_cmp(
2042 labels: &[String],
2043 a: &Value<'_>,
2044 b: &Value<'_>,
2045) -> Option<core::cmp::Ordering> {
2046 let pos = |v: &Value<'_>| -> Option<usize> {
2047 match v {
2048 Value::Text(s) => labels.iter().position(|l| l.as_str() == s.as_ref()),
2049 _ => None,
2050 }
2051 };
2052 Some(pos(a)?.cmp(&pos(b)?))
2053}
2054
2055/// v7.39 (enum order knife) — Binary-comparison hook: when either side's
2056/// static type witnesses an enum and both runtime values are member labels,
2057/// compare by member order (PG's enumsortorder semantics). Out-of-line to
2058/// keep `eval_expr`'s recursion frame small.
2059#[inline(never)]
2060fn enum_compare_hook(
2061 op: BinOp,
2062 lhs: &Expr,
2063 rhs: &Expr,
2064 l: &Value<'_>,
2065 r: &Value<'_>,
2066 ctx: &EvalContext<'_>,
2067) -> Option<Result<Value<'static>, EvalError>> {
2068 let cat = ctx.catalog?;
2069 if cat.enum_types().is_empty() {
2070 return None;
2071 }
2072 let labels = expr_enum_labels(lhs, ctx.columns, ctx.catalog)
2073 .or_else(|| expr_enum_labels(rhs, ctx.columns, ctx.catalog))?;
2074 let ord = enum_ord_cmp(labels, l, r)?;
2075 let b = match op {
2076 BinOp::Eq => ord == core::cmp::Ordering::Equal,
2077 BinOp::NotEq => ord != core::cmp::Ordering::Equal,
2078 BinOp::Lt => ord == core::cmp::Ordering::Less,
2079 BinOp::LtEq => ord != core::cmp::Ordering::Greater,
2080 BinOp::Gt => ord == core::cmp::Ordering::Greater,
2081 BinOp::GtEq => ord != core::cmp::Ordering::Less,
2082 _ => return None,
2083 };
2084 Some(Ok(Value::Bool(b)))
2085}
2086
2087/// v7.39 (round 693) — Binary-comparison hook for a declared collation, the
2088/// last shape F36 left open: `loc BETWEEN 'a' AND 'd'` returns a different
2089/// ROW SET under `en_US.utf8` than under byte order, not merely a different
2090/// order.
2091///
2092/// It sits beside [`enum_compare_hook`] because it is the same kind of fact
2093/// — something about the operand COLUMNS that `compare` cannot look up from
2094/// two values — and takes the same two protections: `#[inline(never)]`, so
2095/// `eval_expr`'s recursion frame does not grow (the comment at the call site
2096/// records a fourth `||` there tipping the 768 KiB guard on its own), and
2097/// the caller's Text/Text gate, so no integer comparison reaches it.
2098///
2099/// EQUALITY is deliberately not handled. PG18's `en_US.utf8` is
2100/// deterministic, so `=`, `<>`, `LIKE`, `IN` and `count(DISTINCT …)` give
2101/// byte-equality's answer — measured, all five. Only the ordering operators
2102/// change, and `least`/`greatest` follow them through their own comparator.
2103#[inline(never)]
2104fn collate_compare_hook(
2105 op: BinOp,
2106 lhs: &Expr,
2107 rhs: &Expr,
2108 l: &Value<'_>,
2109 r: &Value<'_>,
2110 ctx: &EvalContext<'_>,
2111) -> Option<Result<Value<'static>, EvalError>> {
2112 if !matches!(op, BinOp::Lt | BinOp::LtEq | BinOp::Gt | BinOp::GtEq) {
2113 return None;
2114 }
2115 let (Value::Text(a), Value::Text(b)) = (l, r) else {
2116 return None;
2117 };
2118 let resolve = |c: &spg_sql::ast::ColumnName| -> Option<alloc::string::String> {
2119 let pos = find_column_pos(c, ctx)?;
2120 ctx.columns.get(pos)?.collation_name.clone()
2121 };
2122 let derived = crate::collate_derive::derive(lhs, &resolve)
2123 .combine_pub(crate::collate_derive::derive(rhs, &resolve));
2124 if let Some((x, y)) = derived.conflict() {
2125 return Some(Err(EvalError::TypeMismatch {
2126 detail: alloc::format!(
2127 "collation mismatch between implicit collations \"{x}\" and \"{y}\""
2128 ),
2129 }));
2130 }
2131 let ord = crate::collate::compare(derived.name()?, a, b)?;
2132 let b = match op {
2133 BinOp::Lt => ord == core::cmp::Ordering::Less,
2134 BinOp::LtEq => ord != core::cmp::Ordering::Greater,
2135 BinOp::Gt => ord == core::cmp::Ordering::Greater,
2136 BinOp::GtEq => ord != core::cmp::Ordering::Less,
2137 _ => return None,
2138 };
2139 Some(Ok(Value::Bool(b)))
2140}
2141
2142/// v7.39 (round 693) — the collation `least`/`greatest` should compare by,
2143/// derived across every argument the same way a comparison's two operands
2144/// are. `None` keeps byte order, which is right for arguments that declare
2145/// nothing.
2146#[inline(never)]
2147fn greatest_least_collation(args: &[Expr], ctx: &EvalContext<'_>) -> Option<alloc::string::String> {
2148 let resolve = |c: &spg_sql::ast::ColumnName| -> Option<alloc::string::String> {
2149 let pos = find_column_pos(c, ctx)?;
2150 ctx.columns.get(pos)?.collation_name.clone()
2151 };
2152 let derived = args
2153 .iter()
2154 .fold(crate::collate_derive::Derived::None, |acc, a| {
2155 acc.combine_pub(crate::collate_derive::derive(a, &resolve))
2156 });
2157 derived
2158 .name()
2159 .filter(|n| crate::collate::is_supported(n))
2160 .map(alloc::string::ToString::to_string)
2161}
2162
2163/// v7.39 (round 704) — rewrite a comparison's operator-not-found error when
2164/// the operand at fault is an UNKNOWN string literal against a numeric-family
2165/// value. PG commits such a literal to the other side's type before comparing,
2166/// so its error is the input function's — `invalid input syntax for type
2167/// integer: "abc"` — not `operator does not exist: integer = text`. An
2168/// explicit `::text` operand keeps the operator error (`1 IS DISTINCT FROM
2169/// 'a'::text`, measured on PG18), which is precisely the distinction two
2170/// `Value`s cannot carry: the first cut of this round rewrote inside
2171/// `binop::compare` and the r238 pin plus corpus 19 caught it the same day.
2172///
2173/// Error-path only — a comparison that succeeds never calls this — so the
2174/// 35.6 %-of-self-time note on `compare` is untouched.
2175#[cold]
2176#[inline(never)]
2177fn unknown_literal_cmp_error(
2178 err: EvalError,
2179 lhs: &Expr,
2180 rhs: &Expr,
2181 lv: &Value<'_>,
2182 rv: &Value<'_>,
2183) -> EvalError {
2184 let EvalError::TypeMismatch { detail } = &err else {
2185 return err;
2186 };
2187 // Two spellings of the same fall-through: `compare`'s operator error,
2188 // and the owned numeric path's conversion error (`f = 'y'` reaches
2189 // "cannot convert text to FLOAT"). Both mean the literal failed to
2190 // lift; neither is what PG says about an unknown literal.
2191 if !detail.starts_with("operator does not exist")
2192 && !detail.starts_with("cannot convert text to")
2193 {
2194 return err;
2195 }
2196 let numeric = |v: &Value<'_>| {
2197 matches!(
2198 v.data_type(),
2199 Some(
2200 spg_storage::DataType::SmallInt
2201 | spg_storage::DataType::Int
2202 | spg_storage::DataType::BigInt
2203 | spg_storage::DataType::Float
2204 | spg_storage::DataType::Real
2205 | spg_storage::DataType::Numeric { .. }
2206 )
2207 )
2208 };
2209 let rewrite = |s: &Value<'_>, other: &Value<'_>| -> Option<EvalError> {
2210 let Value::Text(text) = s else { return None };
2211 let dt = other.data_type()?;
2212 Some(EvalError::TypeMismatch {
2213 detail: alloc::format!(
2214 "invalid input syntax for type {}: \"{text}\"",
2215 crate::conversions::pg_type_name_for_error(dt)
2216 ),
2217 })
2218 };
2219 if is_unknown_string_literal(lhs)
2220 && numeric(rv)
2221 && let Some(e) = rewrite(lv, rv)
2222 {
2223 return e;
2224 }
2225 if is_unknown_string_literal(rhs)
2226 && numeric(lv)
2227 && let Some(e) = rewrite(rv, lv)
2228 {
2229 return e;
2230 }
2231 err
2232}
2233
2234fn enum_arg_type_name<'e>(args: &'e [Expr], ctx: &EvalContext<'e>) -> Option<&'e str> {
2235 args.iter()
2236 .find_map(|a| expr_enum_type_name(a, ctx.columns))
2237 .filter(|n| {
2238 ctx.catalog
2239 .is_some_and(|cat| cat.enum_types().contains_key(*n))
2240 })
2241}
2242
2243/// Cheap value-free precheck so `eval_expr`'s recursion frame carries no
2244/// binding for the enum path (stack-depth guard budget).
2245#[inline(never)]
2246fn enum_introspection_applies(args: &[Expr], ctx: &EvalContext<'_>) -> bool {
2247 enum_arg_type_name(args, ctx).is_some()
2248}
2249
2250#[inline(never)]
2251fn eval_enum_introspection(
2252 name: &str,
2253 args: &[Expr],
2254 row: &Row<'static>,
2255 ctx: &EvalContext<'_>,
2256) -> Result<Value<'static>, EvalError> {
2257 let Some(en) = enum_arg_type_name(args, ctx)
2258 .and_then(|n| ctx.catalog.and_then(|cat| cat.enum_types().get(n)))
2259 else {
2260 // The precheck guarantees this arm is unreachable; keep a typed
2261 // error rather than a panic if the two ever drift.
2262 return Err(EvalError::TypeMismatch {
2263 detail: "could not determine polymorphic type".into(),
2264 });
2265 };
2266 let labels = &en.labels;
2267 if labels.is_empty() {
2268 return Ok(Value::Null);
2269 }
2270 if name.eq_ignore_ascii_case("enum_first") {
2271 return Ok(Value::text(labels[0].clone()));
2272 }
2273 if name.eq_ignore_ascii_case("enum_last") {
2274 return Ok(Value::text(labels[labels.len() - 1].clone()));
2275 }
2276 // enum_range(NULL) = all; enum_range(lo, hi) slices inclusively,
2277 // NULL bound = open end (PG).
2278 let pos_of = |v: &Value<'_>| -> Option<usize> {
2279 match v {
2280 Value::Text(s) => labels.iter().position(|l| l == s.as_ref()),
2281 _ => None,
2282 }
2283 };
2284 let (lo, hi) = if args.len() == 2 {
2285 let a = eval_expr(&args[0], row, ctx)?;
2286 let b = eval_expr(&args[1], row, ctx)?;
2287 (
2288 pos_of(&a).unwrap_or(0),
2289 pos_of(&b).unwrap_or(labels.len() - 1),
2290 )
2291 } else {
2292 (0, labels.len() - 1)
2293 };
2294 let out: alloc::vec::Vec<Option<String>> = labels
2295 .get(lo..=hi)
2296 .unwrap_or(&[])
2297 .iter()
2298 .map(|l| Some(l.clone()))
2299 .collect();
2300 Ok(Value::TextArray(out))
2301}
2302
2303/// Apply one `[index]` subscript to a value — the single-step semantics shared
2304/// by 1-D array elements and JSON path access (`j['a']`, `j[0]`). NULL target
2305/// or index → NULL; a 1-based integer indexes a 1-D array (out of range → NULL,
2306/// non-array → error); JSON delegates to `path_get`.
2307fn apply_one_subscript(
2308 target_v: Value<'static>,
2309 index: &Expr,
2310 row: &Row<'static>,
2311 ctx: &EvalContext<'_>,
2312) -> Result<Value<'static>, EvalError> {
2313 let idx_v = eval_expr(index, row, ctx)?;
2314 if matches!(target_v, Value::Null) || matches!(idx_v, Value::Null) {
2315 return Ok(Value::Null);
2316 }
2317 // v7.38 (read01) — JSON/JSONB subscripting (`j['a']`, `j[0]`, chained
2318 // `j['a']['b']`) is object/array access, identical to the `->` operator
2319 // (text key → object field, integer → 0-based array element). PG 14+.
2320 if matches!(target_v, Value::Json(_)) {
2321 return crate::json::path_get(&target_v, &idx_v, false);
2322 }
2323 let i: i64 = match idx_v {
2324 Value::Int(n) => i64::from(n),
2325 Value::BigInt(n) => n,
2326 Value::SmallInt(n) => i64::from(n),
2327 other => {
2328 return Err(EvalError::TypeMismatch {
2329 detail: format!(
2330 "array subscript must be integer, got {}",
2331 crate::conversions::pg_type_name_for_error_opt(other.data_type())
2332 ),
2333 });
2334 }
2335 };
2336 if i < 1 {
2337 return Ok(Value::Null);
2338 }
2339 let pos = (i - 1) as usize;
2340 match array_element_at(&target_v, pos) {
2341 Some(v) => Ok(v),
2342 None if array_len(&target_v).is_some() => Ok(Value::Null),
2343 None => Err(EvalError::TypeMismatch {
2344 detail: format!(
2345 "subscript target must be an array, got {}",
2346 crate::conversions::pg_type_name_for_error_opt(target_v.data_type())
2347 ),
2348 }),
2349 }
2350}
2351
2352/// v7.38 (read01, 2D-subscript) — index a 2-D array (`arr[i][j]`). PG needs
2353/// exactly two subscripts to reach an element; a single subscript on a 2-D
2354/// array yields NULL (not the row), and any out-of-range index → NULL. Both
2355/// subscripts are 1-based.
2356fn eval_matrix_subscript(
2357 base: &Value<'static>,
2358 idx_exprs: &[&Expr],
2359 row: &Row<'static>,
2360 ctx: &EvalContext<'_>,
2361) -> Result<Value<'static>, EvalError> {
2362 if idx_exprs.len() != 2 {
2363 return Ok(Value::Null);
2364 }
2365 let mut idx = [0i64; 2];
2366 for (k, ix) in idx_exprs.iter().enumerate() {
2367 idx[k] = match eval_expr(ix, row, ctx)? {
2368 Value::Null => return Ok(Value::Null),
2369 Value::Int(n) => i64::from(n),
2370 Value::BigInt(n) => n,
2371 Value::SmallInt(n) => i64::from(n),
2372 other => {
2373 return Err(EvalError::TypeMismatch {
2374 detail: format!(
2375 "array subscript must be integer, got {}",
2376 crate::conversions::pg_type_name_for_error_opt(other.data_type())
2377 ),
2378 });
2379 }
2380 };
2381 }
2382 let (r, c) = (idx[0], idx[1]);
2383 if r < 1 || c < 1 {
2384 return Ok(Value::Null);
2385 }
2386 let (ri, ci) = ((r - 1) as usize, (c - 1) as usize);
2387 macro_rules! elem {
2388 ($rows:expr, $map:expr) => {
2389 Ok($rows
2390 .get(ri)
2391 .and_then(|inner| inner.get(ci))
2392 .map_or(Value::Null, |cell| cell.as_ref().map_or(Value::Null, $map)))
2393 };
2394 }
2395 match base {
2396 Value::IntArray2D(rows) => elem!(rows, |n| Value::Int(*n)),
2397 Value::BigIntArray2D(rows) => elem!(rows, |n| Value::BigInt(*n)),
2398 Value::BoolArray2D(rows) => elem!(rows, |b| Value::Bool(*b)),
2399 Value::TextArray2D(rows) => {
2400 elem!(rows, |s| Value::Text(alloc::borrow::Cow::Owned(s.clone())))
2401 }
2402 _ => Ok(Value::Null),
2403 }
2404}
2405
2406/// Out-of-lined `eval_expr` arm — keeps the recursive frame small
2407/// (stack-depth guard budget); body unchanged.
2408#[inline(never)]
2409fn eval_cast_arm(
2410 expr: &Expr,
2411 target: &CastTarget,
2412 row: &Row<'static>,
2413 ctx: &EvalContext<'_>,
2414) -> Result<Value<'static>, EvalError> {
2415 let v = eval_expr(expr, row, ctx)?;
2416 // v7.39 (round 473) — `<oid>::regclass` names the relation.
2417 //
2418 // The cast itself has no catalog, so it answered the bare number:
2419 // `indexrelid::regclass` printed `100001` where PG prints `ix1`, and
2420 // pg_class / pg_index rows are read by joining on oids and rendering
2421 // them — a tool cannot match the two up. `relation_name_for_oid`
2422 // mirrors `relation_oid`'s walks so the two directions agree; an oid
2423 // that names nothing keeps rendering as the number, which is what PG
2424 // does for a dropped relation's oid too.
2425 if matches!(target, CastTarget::RegClass)
2426 && let Some(cat) = ctx.catalog
2427 {
2428 let oid = match &v {
2429 Value::Int(n) => Some(i64::from(*n)),
2430 Value::BigInt(n) => Some(*n),
2431 _ => None,
2432 };
2433 if let Some(oid) = oid
2434 && let Some(name) = crate::system_catalog::relation_name_for_oid(cat, oid)
2435 {
2436 return Ok(Value::text(name));
2437 }
2438 }
2439 // v7.38 (read01 P6.40) — a cast to a user DOMAIN (`x::posint`)
2440 // enforces the domain's NOT NULL + CHECK constraints, matching PG.
2441 // The base-type coercion already happened when `v` was produced
2442 // (the domain is a constrained alias of its base type); here we
2443 // only run the constraints.
2444 if let CastTarget::Named(name) = target
2445 && let Some(cat) = ctx.catalog
2446 {
2447 if let Some(dom) = cat.domain_types().get(name.as_str()) {
2448 return apply_domain_constraints(v, dom, name, cat);
2449 }
2450 // v7.38 (read01 P6.67) — `'label'::<user enum>` validates the
2451 // label against the enum's members (a non-member errors like
2452 // PG). A typed NULL passes through carrying the enum type.
2453 if let Some(en) = cat.enum_types().get(name.as_str()) {
2454 return apply_enum_cast(v, en, name);
2455 }
2456 // v7.39 (read01 rowtypes.c) — `'(1,x)'::<composite>` parses PG's
2457 // record text form against the type's field list; a ROW value
2458 // re-labels its fields.
2459 if let Some(comp) = cat.composite_types().get(name.as_str()) {
2460 // v7.39 (round 264) — pass the catalog so a NESTED composite
2461 // field resolves into a record rather than staying text.
2462 return apply_composite_cast_in(v, comp, ctx.catalog);
2463 }
2464 // v7.39 (round 509) — every TABLE also names a row type, and
2465 // `jsonb_populate_record(NULL::mytable, …)` is PG's canonical
2466 // spelling for "shaped like this table". PG accepts `NULL::mytable`
2467 // and refuses `1::mytable` with "cannot cast type integer to
2468 // mytable" — the type exists, the conversion does not. This only
2469 // ever worked here because a NULL skipped cast resolution entirely;
2470 // now that it does not, the row type has to be named explicitly.
2471 if cat.get(name.as_str()).is_some() {
2472 return if matches!(v, Value::Null) {
2473 Ok(Value::Null)
2474 } else {
2475 Err(EvalError::TypeMismatch {
2476 detail: alloc::format!(
2477 "cannot cast type {} to {name}",
2478 crate::eval::strings::pg_typeof_name(&v),
2479 ),
2480 })
2481 };
2482 }
2483 // v7.39 (round 513) — `regnamespace` and `regrole` resolve against
2484 // things that live outside the type table: schemas on the catalog,
2485 // roles on the engine. They belong here for the same reason the
2486 // relation check above does — this is the arm that can see them.
2487 // v7.39 (round 526) — the NUMERIC direction, which is the one a
2488 // catalog join uses: `relnamespace::regnamespace` names the
2489 // schema a relation lives in, and it errored with "unsupported
2490 // cast target" while `'public'::regnamespace` worked. Round 513
2491 // added the name direction only, so the half that reads a
2492 // catalog was the half missing.
2493 if (name.eq_ignore_ascii_case("regnamespace") || name.eq_ignore_ascii_case("regrole"))
2494 && let Some(oid) = match &v {
2495 Value::Int(n) => Some(i64::from(*n)),
2496 Value::BigInt(n) => Some(*n),
2497 _ => None,
2498 }
2499 {
2500 let named = if name.eq_ignore_ascii_case("regnamespace") {
2501 crate::system_catalog::schema_name_for_oid(oid)
2502 } else {
2503 ctx.engine.and_then(|e| e.role_name_for_oid(oid))
2504 };
2505 // PG prints the bare number for an oid that names nothing,
2506 // exactly as `regclass` does.
2507 return Ok(Value::text(
2508 named.unwrap_or_else(|| alloc::format!("{oid}")),
2509 ));
2510 }
2511 if name.eq_ignore_ascii_case("regnamespace")
2512 && let Value::Text(t) = &v
2513 {
2514 let want = t.trim().trim_matches('"');
2515 return if spg_storage::is_builtin_schema(want) || cat.schema_exists(want) {
2516 Ok(Value::text(want.to_string()))
2517 } else {
2518 Err(EvalError::TypeMismatch {
2519 detail: alloc::format!("schema \"{want}\" does not exist"),
2520 })
2521 };
2522 }
2523 if name.eq_ignore_ascii_case("regrole")
2524 && let Value::Text(t) = &v
2525 {
2526 let want = t.trim().trim_matches('"').to_string();
2527 // PG ships predefined roles that exist whether or not anybody
2528 // created them; SPG carries the rest on the engine.
2529 const PREDEFINED: &[&str] = &[
2530 "pg_read_all_data",
2531 "pg_write_all_data",
2532 "pg_monitor",
2533 "pg_read_all_settings",
2534 "pg_read_all_stats",
2535 "pg_stat_scan_tables",
2536 "pg_signal_backend",
2537 "pg_checkpoint",
2538 "pg_maintain",
2539 "pg_use_reserved_connections",
2540 "pg_create_subscription",
2541 ];
2542 let known = PREDEFINED.iter().any(|r| r.eq_ignore_ascii_case(&want))
2543 || ctx.engine.is_some_and(|e| e.role_exists(&want));
2544 return if known {
2545 Ok(Value::text(want))
2546 } else {
2547 Err(EvalError::TypeMismatch {
2548 detail: alloc::format!("role \"{want}\" does not exist"),
2549 })
2550 };
2551 }
2552 // v7.39 (round 509) — the cast target is checked even when the
2553 // operand is NULL. `cast_value_in` short-circuits a NULL before it
2554 // looks at the target, so `NULL::nosuchtype` silently answered NULL
2555 // and `pg_typeof(NULL::nosuchtype)` answered `unknown`, while
2556 // `1::nosuchtype` errored — the gap was exactly the NULL case, in
2557 // both spellings. Everything a catalog can name has been tried by
2558 // now; what is left is the builtin table.
2559 if matches!(v, Value::Null)
2560 && !crate::eval::cast::builtin_target_resolves(name, ctx.mysql_dialect)
2561 {
2562 return Err(EvalError::TypeMismatch {
2563 detail: cast::unknown_type_error_text(name),
2564 });
2565 }
2566 }
2567 // v7.39 (round 285) — `::record`, the anonymous composite type. PG
2568 // treats it as an IDENTITY cast on anything already composite: the
2569 // value keeps its fields and their names, so `(ROW(1,2)::record).f1`
2570 // and `(r).x` still resolve. Only a non-composite is refused, with
2571 // PG's wording. `record` is not a catalog type, so this cannot live
2572 // in the lookups above.
2573 if let CastTarget::Named(name) = target
2574 && name.eq_ignore_ascii_case("record")
2575 {
2576 return match v {
2577 Value::Composite(_) | Value::Null => Ok(v),
2578 other => Err(EvalError::TypeMismatch {
2579 detail: alloc::format!(
2580 "cannot cast type {} to record",
2581 crate::eval::strings::pg_typeof_name(&other),
2582 ),
2583 }),
2584 };
2585 }
2586 // v7.38 (read01, T22) — a numeric OID cast to regclass reverse-looks
2587 // up the user relation name (PG's 16384+ band, assigned in
2588 // table_names() order). System OIDs / non-matches fall through to
2589 // the integer-rendering path in cast_value.
2590 if matches!(target, CastTarget::RegClass) {
2591 // v7.39 (read01 ruleutils.c) — regclass is DUAL-shape: oid for
2592 // catalog joins (conrelid = 't'::regclass), name for display.
2593 let oid_in = match &v {
2594 Value::SmallInt(n) => Some(i64::from(*n)),
2595 Value::Int(n) => Some(i64::from(*n)),
2596 Value::BigInt(n) => Some(*n),
2597 _ => None,
2598 };
2599 if let (Some(oid), Some(cat)) = (oid_in, ctx.catalog) {
2600 if oid >= 16384 {
2601 if let Some(name) = cat.table_names().into_iter().nth((oid - 16384) as usize) {
2602 return Ok(Value::RegClass(oid, name.into()));
2603 }
2604 }
2605 }
2606 if let (Value::Text(s), Some(cat)) = (&v, ctx.catalog) {
2607 let bare = s
2608 .rsplit('.')
2609 .next()
2610 .unwrap_or(s)
2611 .trim_matches('"')
2612 .to_string();
2613 if let Some(oid) = regclass_name_to_oid(cat, &bare) {
2614 return Ok(Value::RegClass(oid, bare.into()));
2615 }
2616 // v7.39 (round 337, V62) — a name that is no relation at all is
2617 // PG's error, not a silent pass-through. `'nope'::regclass`
2618 // used to answer the TEXT `nope`, so a downstream
2619 // `pg_get_viewdef('nope'::regclass)` reported "no such view"
2620 // when the truth is there is no such relation — and a
2621 // catalog join on it quietly matched nothing. PG 18.4:
2622 // `ERROR: relation "nope" does not exist`. (`to_regclass` is
2623 // the spelling that answers NULL instead, and still does.)
2624 //
2625 // The system views SPG synthesises have no oid space, so they
2626 // keep the textual form rather than erroring.
2627 const SYSTEM_RELS: &[&str] = &[
2628 "pg_roles",
2629 "pg_user",
2630 "pg_tables",
2631 "pg_views",
2632 "pg_settings",
2633 "pg_stat_activity",
2634 "pg_stat_database",
2635 "pg_stat_user_tables",
2636 "pg_class",
2637 "pg_attribute",
2638 "pg_type",
2639 "pg_proc",
2640 "pg_namespace",
2641 "pg_constraint",
2642 "pg_index",
2643 "pg_rewrite",
2644 ];
2645 if !SYSTEM_RELS.contains(&bare.as_str()) {
2646 return Err(EvalError::TypeMismatch {
2647 detail: alloc::format!("relation \"{bare}\" does not exist"),
2648 });
2649 }
2650 }
2651 }
2652 // v7.39 (round 339, V63) — `::regproc` / `::regprocedure` resolve
2653 // against the USER function catalog too. Name resolution ran against
2654 // the static pg_proc table alone, so `'my_fn'::regproc` — the form
2655 // every catalog query and pg_dump uses to name a function — raised
2656 // `function "my_fn" does not exist` for a function that plainly did.
2657 // The cast layer has no catalog handle; this is the same interception
2658 // point the `::regclass` block above uses.
2659 if let (CastTarget::Named(tname), Some(cat), Value::Text(s)) = (target, ctx.catalog, &v) {
2660 let lower = tname.to_ascii_lowercase();
2661 if matches!(lower.as_str(), "regproc" | "regprocedure") {
2662 let raw = s.trim();
2663 // regprocedure carries the argument list: `f(int,text)`.
2664 let (name_part, args_part) = match raw.split_once('(') {
2665 Some((n, rest)) => (n.trim(), Some(rest.trim_end_matches(')'))),
2666 None => (raw, None),
2667 };
2668 let bare = name_part
2669 .strip_prefix("public.")
2670 .unwrap_or(name_part)
2671 .trim_matches('"');
2672 let cands = cat.functions_named(bare);
2673 if let Some(args_txt) = args_part {
2674 // An overload IS distinguishable here — the argument list
2675 // is what regprocedure exists to carry.
2676 let want =
2677 crate::system_catalog::canonical_arg_types(&alloc::format!("({args_txt})"));
2678 if let Some(f) = cands
2679 .iter()
2680 .find(|f| crate::system_catalog::canonical_arg_types(&f.args_repr) == want)
2681 {
2682 let rendered = alloc::format!(
2683 "{bare}({})",
2684 crate::system_catalog::canonical_arg_types(&f.args_repr)
2685 );
2686 // v7.39 (round 342, V65) — dual shape: the oid for
2687 // catalog joins, the rendering for display.
2688 let oid = crate::system_catalog::function_oid_by_signature(cat, bare, &want)
2689 .unwrap_or(0);
2690 return Ok(Value::RegProc(oid, rendered.into()));
2691 }
2692 } else {
2693 match cands.len() {
2694 0 => {}
2695 1 => {
2696 let oid = crate::system_catalog::function_oid(cat, bare).unwrap_or(0);
2697 return Ok(Value::RegProc(oid, bare.into()));
2698 }
2699 _ => {
2700 return Err(EvalError::TypeMismatch {
2701 detail: alloc::format!("more than one function named \"{bare}\""),
2702 });
2703 }
2704 }
2705 }
2706 }
2707 }
2708 // v7.38 (T-tstz Phase 1) — `<timestamptz>::text` renders the offset
2709 // (`2024-01-15 10:30:00+00`); plain timestamp does not. The runtime
2710 // value is the same tz-less `Value::Timestamp`, so consult the
2711 // inner expression's static type. Falls through to the ordinary
2712 // cast on any shape the static typer can't resolve — worst case is
2713 // today's no-offset rendering, never a wrong instant.
2714 if matches!(target, CastTarget::Text)
2715 && let Value::Timestamp(t) = &v
2716 && crate::describe::describe_expr(expr, ctx.columns)
2717 .is_some_and(|s| matches!(s.ty, spg_storage::DataType::Timestamptz))
2718 {
2719 // v7.39 (tz epic) — per-VALUE session offset (DST zones
2720 // vary within a statement); non-ISO DateStyles carry the
2721 // zone designation instead of the numeric offset.
2722 let off = ctx.session_tz_offset_at(*t);
2723 let abbr = ctx.session_tz_abbrev_at(*t);
2724 return Ok(Value::text(format::format_timestamptz_tz(
2725 *t,
2726 &ctx.render_style,
2727 off,
2728 abbr.as_deref(),
2729 )));
2730 }
2731 // v7.39 (read01 utils/adt, datetime.c) — the relative
2732 // reserved words resolve against the transaction clock:
2733 // 'today'/'tomorrow'/'yesterday' are midnight dates, 'now'
2734 // is the current instant ('now'::date = today). Clockless
2735 // engines fall through to the parser (which rejects them).
2736 if matches!(
2737 target,
2738 CastTarget::Date | CastTarget::Timestamp | CastTarget::Timestamptz
2739 ) && let Value::Text(word) = &v
2740 && let Some(clock) = ctx.clock
2741 {
2742 let w = word.trim().to_ascii_lowercase();
2743 if matches!(w.as_str(), "today" | "tomorrow" | "yesterday" | "now") {
2744 let now_us = clock();
2745 let today = i32::try_from(now_us.div_euclid(86_400_000_000)).ok();
2746 if let Some(today) = today {
2747 let day = match w.as_str() {
2748 "tomorrow" => today + 1,
2749 "yesterday" => today - 1,
2750 _ => today,
2751 };
2752 return Ok(match (&target, w.as_str()) {
2753 (CastTarget::Date, _) => Value::Date(day),
2754 (_, "now") => Value::Timestamp(now_us),
2755 _ => Value::Timestamp(crate::conversions::date_days_to_micros(day)),
2756 });
2757 }
2758 }
2759 }
2760 // v7.39 (GUC knife 5) — text INPUT to date/timestamp under a
2761 // non-MDY DateOrder disambiguates by the session order
2762 // (`'01/02/2024'::date` is Feb 1 under DMY). The default MDY
2763 // order flows through cast_value's parse_date_literal.
2764 if ctx.render_style.date_order != format::DateOrder::Mdy {
2765 match (&target, &v) {
2766 (CastTarget::Date, Value::Text(s)) => {
2767 if let Some(d) = format::parse_date_literal_ordered(s, ctx.render_style.date_order)
2768 {
2769 return Ok(Value::Date(d));
2770 }
2771 }
2772 (CastTarget::Timestamp, Value::Text(s)) => {
2773 if let Some(t) =
2774 format::parse_timestamp_literal_ordered(s, ctx.render_style.date_order)
2775 {
2776 return Ok(Value::Timestamp(t));
2777 }
2778 }
2779 _ => {}
2780 }
2781 }
2782 // v7.39 (round 309, V30) — the mirror of the timestamptz arm
2783 // below. A literal carrying a zone NAME is legal input to the
2784 // zone-less types, and PG throws the zone away rather than
2785 // converting: `'2020-01-01 10:00:00 America/New_York'::timestamp`
2786 // is 10:00, not 15:00. Round 289 did this for a numeric `+02`
2787 // offset; a named zone still failed to parse at all.
2788 //
2789 // The name is not simply stripped — PG validates it, and says so
2790 // (`time zone "bogus/zone" not recognized`, lowercased) rather than
2791 // reporting a malformed literal. That check is why this belongs
2792 // here and not in `cast_value`: resolving a zone needs the host
2793 // functions, which only the context carries.
2794 let zoneless_target = match &target {
2795 CastTarget::Timestamp => Some("timestamp"),
2796 CastTarget::Date => Some("date"),
2797 // `::time` has no CastTarget of its own; it arrives named.
2798 CastTarget::Named(n) if n.eq_ignore_ascii_case("time") => Some("time"),
2799 _ => None,
2800 };
2801 if let Some(kind) = zoneless_target
2802 && let Value::Text(txt) = &v
2803 && let Some((wall, zone)) = split_trailing_zone_name(txt, ctx.render_style.date_order)
2804 {
2805 if ctx.zone_local_to_utc(zone, wall).is_none() {
2806 // Measured boundary: PG calls the token a ZONE NAME — and
2807 // so reports a misspelling as such — only when it is
2808 // path-shaped. A bare word it does not know (`ABCD`, `QQQ`,
2809 // `UTC_X`) makes the whole literal invalid syntax instead,
2810 // because nothing marks it as having meant a zone at all.
2811 if zone.contains('/') {
2812 return Err(EvalError::TypeMismatch {
2813 detail: alloc::format!(
2814 "time zone \"{}\" not recognized",
2815 zone.to_ascii_lowercase()
2816 ),
2817 });
2818 }
2819 } else {
2820 return Ok(match kind {
2821 "date" => {
2822 Value::Date(i32::try_from(wall.div_euclid(86_400_000_000)).map_err(|_| {
2823 EvalError::TypeMismatch {
2824 detail: "timestamp out of DATE range".into(),
2825 }
2826 })?)
2827 }
2828 "time" => Value::Time(wall.rem_euclid(86_400_000_000)),
2829 _ => Value::Timestamp(wall),
2830 });
2831 }
2832 }
2833 // v7.39 (tz epic) — timestamptz INPUT: an offset-less
2834 // literal is a wall-clock reading in the session zone
2835 // (PG); a trailing IANA zone name localises there. Both
2836 // fall through to cast_value when nothing matches (its
2837 // parse treats naive input as UTC — correct for a UTC
2838 // session).
2839 if matches!(target, CastTarget::Timestamptz)
2840 && let Value::Text(txt) = &v
2841 {
2842 let order = ctx.render_style.date_order;
2843 let sess_zone = ctx
2844 .session_gucs
2845 .and_then(|g| g.get("timezone"))
2846 .map(String::as_str);
2847 // Trailing zone name: the last space-separated token,
2848 // when it names a resolvable zone (contains a letter
2849 // and isn't consumed by the plain parse).
2850 if let Some(idx) = txt.trim_end().rfind(' ') {
2851 let (head, tail) = (txt[..idx].trim(), txt[idx + 1..].trim());
2852 let tail_is_zoneish = tail.len() > 1
2853 && tail.bytes().any(|b| b.is_ascii_alphabetic())
2854 && !tail.eq_ignore_ascii_case("bc")
2855 && !tail.eq_ignore_ascii_case("ad");
2856 if tail_is_zoneish
2857 && format::parse_timestamp_literal_tz_ordered(txt, order).is_none()
2858 && let Some((wall, false)) = format::parse_timestamp_literal_tz_ordered(head, order)
2859 && let Some(utc) = ctx.zone_local_to_utc(tail, wall)
2860 {
2861 return Ok(Value::Timestamp(utc));
2862 }
2863 }
2864 if let Some((wall, had_tz)) = format::parse_timestamp_literal_tz_ordered(txt, order) {
2865 if had_tz {
2866 return Ok(Value::Timestamp(wall));
2867 }
2868 if let Some(zone) = sess_zone
2869 && !zone.eq_ignore_ascii_case("utc")
2870 && !zone.eq_ignore_ascii_case("gmt")
2871 && let Some(utc) = ctx.zone_local_to_utc(zone, wall)
2872 {
2873 return Ok(Value::Timestamp(utc));
2874 }
2875 return Ok(Value::Timestamp(wall));
2876 }
2877 }
2878 // v7.39 (round 523) — a NAIVE timestamp cast to timestamptz is a
2879 // wall-clock reading in the session zone, exactly as the text form
2880 // above already is. This was a no-op, so under `SET TimeZone =
2881 // 'Asia/Tokyo'` a `TIMESTAMP '2020-01-01 00:00:00'::timestamptz`
2882 // named 09:00 JST — a different INSTANT, nine hours from the one PG
2883 // stores, not a different rendering of the same one.
2884 //
2885 // The source's static type is the witness: SPG keeps timestamptz in
2886 // the same `Value::Timestamp`, so only an expression that is not
2887 // ALREADY timestamptz may be shifted, or a tstz-to-tstz cast would
2888 // move the instant twice.
2889 if matches!(target, CastTarget::Timestamptz)
2890 && let Value::Timestamp(wall) = &v
2891 && !matches!(
2892 crate::describe::describe_expr(expr, ctx.columns).map(|s| s.ty),
2893 Some(spg_storage::DataType::Timestamptz)
2894 )
2895 && let Some(zone) = ctx.session_gucs.and_then(|g| g.get("timezone"))
2896 && !zone.eq_ignore_ascii_case("utc")
2897 && !zone.eq_ignore_ascii_case("gmt")
2898 && let Some(utc) = ctx.zone_local_to_utc(zone, *wall)
2899 {
2900 return Ok(Value::Timestamp(utc));
2901 }
2902 // v7.39 (round 523) — and the other direction: a timestamptz cast
2903 // DOWN to a zone-free type reads the local clock in the session
2904 // zone. `(TIMESTAMPTZ '2020-01-01 15:00:00Z')::date` answered
2905 // 2020-01-01 in Tokyo where PG answers 2020-01-02 — a whole day out
2906 // for every instant in the last nine hours of a UTC day, which is
2907 // exactly the shape a daily report groups on. `now()::timestamp`
2908 // likewise disagreed with `now() AT TIME ZONE <session zone>`, which
2909 // PG defines to be the same value.
2910 if matches!(target, CastTarget::Timestamp | CastTarget::Date)
2911 && let Value::Timestamp(t) = &v
2912 && crate::describe::describe_expr(expr, ctx.columns)
2913 .is_some_and(|s| matches!(s.ty, spg_storage::DataType::Timestamptz))
2914 {
2915 let local = t.saturating_add(ctx.session_tz_offset_at(*t));
2916 return Ok(match target {
2917 CastTarget::Date => i32::try_from(local.div_euclid(86_400_000_000))
2918 .map_or(Value::Timestamp(local), Value::Date),
2919 _ => Value::Timestamp(local),
2920 });
2921 }
2922 // v7.39 (GUC knife 3) — the out-function casts honour the
2923 // session render style, like PG's date_out/interval_out/
2924 // float8out under DateStyle/IntervalStyle/extra_float_digits.
2925 if matches!(target, CastTarget::Text) {
2926 match &v {
2927 // v7.39 (round 524) — `bytea_out` is a render GUC too.
2928 Value::Bytes(b) if ctx.render_style.bytea_escape => {
2929 return Ok(Value::text(format::format_bytea_escape(b)));
2930 }
2931 Value::Date(d) => {
2932 return Ok(Value::text(format::format_date_styled(
2933 *d,
2934 &ctx.render_style,
2935 )));
2936 }
2937 Value::Timestamp(t) => {
2938 return Ok(Value::text(format::format_timestamp_styled(
2939 *t,
2940 &ctx.render_style,
2941 )));
2942 }
2943 Value::Interval {
2944 months,
2945 days,
2946 micros,
2947 } => {
2948 return Ok(Value::text(format::format_interval_styled(
2949 *months,
2950 *days,
2951 *micros,
2952 &ctx.render_style,
2953 )));
2954 }
2955 Value::Float(x) => {
2956 return Ok(Value::text(format::format_float_styled(
2957 *x,
2958 &ctx.render_style,
2959 )));
2960 }
2961 Value::Real(x) => {
2962 return Ok(Value::text(format::format_real_styled(
2963 *x,
2964 &ctx.render_style,
2965 )));
2966 }
2967 _ => {}
2968 }
2969 }
2970 crate::eval::cast::cast_value_ref_in(v, target, ctx.mysql_dialect)
2971}
2972
2973/// Out-of-lined `eval_expr` arm — keeps the recursive frame small
2974/// (stack-depth guard budget); body unchanged.
2975#[inline(never)]
2976fn eval_array_arm(
2977 items: &[Expr],
2978 row: &Row<'static>,
2979 ctx: &EvalContext<'_>,
2980) -> Result<Value<'static>, EvalError> {
2981 let mut materialised: Vec<Value<'static>> = Vec::with_capacity(items.len());
2982 for elem in items {
2983 materialised.push(eval_expr(elem, row, ctx)?);
2984 }
2985 // v7.38 (read01, T10) — a constructor whose elements are all 1-D
2986 // arrays builds a 2-D array (`ARRAY[[1,2],[3,4]]`). All rows must
2987 // share a length (PG: "multidimensional arrays must have array
2988 // expressions with matching dimensions"). Int rows promote to
2989 // bigint if any row is bigint; a text row makes the whole thing text.
2990 // v7.39 (read01 round 73) — a row of ANY array kind counts. Round 72 gave
2991 // `ARRAY[true,false]` its real `bool[]` type, and this detector only knew
2992 // Int / BigInt / Text rows — so `ARRAY[ARRAY[true,false]]` stopped being a
2993 // 2-D array at all and collapsed into a 1-D text[] of rendered rows, with
2994 // `[1][2]` failing outright. A regression THIS campaign introduced, caught by
2995 // the very sweep that was chasing its own residual. Rows that are not
2996 // int/bigint arrays render into the text 2-D form below (SPG has no bool 2-D
2997 // storage variant — a recorded residual), but they stay 2-D.
2998 let all_arrays = !materialised.is_empty()
2999 && materialised.iter().all(|v| {
3000 values::array_len(v).is_some()
3001 && !matches!(
3002 v,
3003 Value::TextArray2D(_)
3004 | Value::IntArray2D(_)
3005 | Value::BigIntArray2D(_)
3006 | Value::BoolArray2D(_)
3007 )
3008 });
3009 if all_arrays {
3010 let row_len = values::array_len(&materialised[0]).unwrap_or(0);
3011 let same_len = materialised
3012 .iter()
3013 .all(|v| values::array_len(v) == Some(row_len));
3014 if !same_len {
3015 return Err(EvalError::TypeMismatch {
3016 detail: "multidimensional arrays must have array expressions \
3017 with matching dimensions"
3018 .into(),
3019 });
3020 }
3021 // v7.39 (read01 round 75) — all-BOOL rows build a real `bool[][]`. BOOL is
3022 // the one element type whose ARRAY rendering (`t`) differs from its
3023 // scalar one (`true`), so the text-backed 2-D could not be right for it:
3024 // `ARRAY[ARRAY[true,false]]::text` wants `{{t,f}}` while `[1][2]::text`
3025 // wants `false`. Every other type renders the same either way — which is
3026 // why this is the only typed 2-D SPG needs.
3027 if materialised
3028 .iter()
3029 .all(|v| matches!(v, Value::BoolArray(_)))
3030 {
3031 let rows: Vec<Vec<Option<bool>>> = materialised
3032 .into_iter()
3033 .map(|v| match v {
3034 Value::BoolArray(r) => r,
3035 _ => unreachable!("checked above"),
3036 })
3037 .collect();
3038 return Ok(Value::BoolArray2D(rows));
3039 }
3040 let any_text = materialised
3041 .iter()
3042 .any(|v| !matches!(v, Value::IntArray(_) | Value::BigIntArray(_)));
3043 let any_big = materialised
3044 .iter()
3045 .any(|v| matches!(v, Value::BigIntArray(_)));
3046 if any_text {
3047 let rows: Vec<Vec<Option<String>>> = materialised
3048 .into_iter()
3049 .map(|v| match v {
3050 Value::TextArray(r) => r,
3051 // Any other element type renders into the text 2-D form,
3052 // element by element. SPG has no typed 2-D storage beyond
3053 // int / bigint / text, so a bool 2-D array IS text — and the
3054 // SCALAR rendering is the one to use: `(arr)[1][2]::text`
3055 // must read `false`, as in PG. (`pg_typeof` reporting
3056 // `text[]` rather than `boolean[]` is the recorded residual;
3057 // a typed 2-D needs new storage variants.)
3058 other => {
3059 let n = values::array_len(&other).unwrap_or(0);
3060 (0..n)
3061 .map(|i| match values::array_element_at(&other, i) {
3062 None | Some(Value::Null) => None,
3063 Some(v) => Some(value_to_text(&v)),
3064 })
3065 .collect()
3066 }
3067 })
3068 .collect();
3069 return Ok(Value::TextArray2D(rows));
3070 }
3071 if any_big {
3072 let rows: Vec<Vec<Option<i64>>> = materialised
3073 .into_iter()
3074 .map(|v| match v {
3075 Value::BigIntArray(r) => r,
3076 Value::IntArray(r) => r.into_iter().map(|c| c.map(i64::from)).collect(),
3077 _ => unreachable!(),
3078 })
3079 .collect();
3080 return Ok(Value::BigIntArray2D(rows));
3081 }
3082 let rows: Vec<Vec<Option<i32>>> = materialised
3083 .into_iter()
3084 .map(|v| match v {
3085 Value::IntArray(r) => r,
3086 _ => unreachable!(),
3087 })
3088 .collect();
3089 return Ok(Value::IntArray2D(rows));
3090 }
3091 // v7.39 (read01 round 72) — a HOMOGENEOUS array of a non-numeric, non-text
3092 // type keeps that type, and is unambiguous, so it is decided BEFORE the
3093 // numeric/text unification below. Everything outside the numeric ladder and
3094 // text used to fall into that loop's `_ => has_text = true` — a silent
3095 // degradation, not a decision: `ARRAY[true, false]` came back as `text[]`.
3096 // It usually LOOKED right (array_to_string renders `t` either way), which is
3097 // exactly what let it sit; the array FUNCTIONS are what tripped over it.
3098 if let Some(v) = values::homogeneous_typed_array(&materialised) {
3099 return Ok(crate::describe::upgrade_timestamptz_array(
3100 v,
3101 items,
3102 ctx.columns,
3103 ));
3104 }
3105 // v7.39 (round 236) — PG resolves an ARRAY constructor's elements to ONE
3106 // element type and refuses the constructor when they have no common one:
3107 // `ARRAY[1, 'a'::text]` is "ARRAY types integer and text cannot be
3108 // matched". SPG degraded to `text[]` instead, so `ARRAY[1, true]` came
3109 // back as `{1,t}` — a column of rendered strings that then behaved like
3110 // text everywhere downstream. Same rule (and the same untyped-literal
3111 // subtlety) as the set-operation resolution in round 233: a bare string
3112 // literal is PG's `unknown` and takes the other elements' type, so it is
3113 // identified from the SYNTAX, not from the value's runtime type.
3114 unify_array_elements(items, &mut materialised)?;
3115 // Coercing the untyped elements can make the array homogeneous
3116 // (`ARRAY[true,'t']` becomes two booleans), so re-try the typed-array
3117 // path before falling into the numeric/text ladder below — otherwise
3118 // the now-uniform boolean array would still degrade to text[].
3119 if let Some(v) = values::homogeneous_typed_array(&materialised) {
3120 return Ok(crate::describe::upgrade_timestamptz_array(
3121 v,
3122 items,
3123 ctx.columns,
3124 ));
3125 }
3126 let mut has_text = false;
3127 let mut has_float = false;
3128 let mut has_numeric = false;
3129 let mut has_bigint = false;
3130 let mut has_int = false;
3131 // A NumericBig or non-finite (NaN/Inf) numeric can't be held in
3132 // NumericArray's `(i128, scale)` cells, so it forces the text[]
3133 // fallback rather than a lossy/panicking conversion.
3134 let mut numeric_representable = true;
3135 for v in &materialised {
3136 match v {
3137 Value::Null => {}
3138 Value::Int(_) | Value::SmallInt(_) => has_int = true,
3139 Value::BigInt(_) => has_bigint = true,
3140 Value::Numeric {
3141 kind: spg_storage::NumericKind::Finite,
3142 ..
3143 } => {
3144 has_numeric = true;
3145 }
3146 Value::Numeric { .. } => {
3147 has_numeric = true;
3148 numeric_representable = false;
3149 }
3150 Value::NumericBig(_) => {
3151 has_numeric = true;
3152 numeric_representable = false;
3153 }
3154 Value::Float(_) => has_float = true,
3155 Value::Text(_) | Value::Json(_) => has_text = true,
3156 // v7.39 (round 652) — a reg value belongs to the array by
3157 // its OID half. Falling into the catch-all made
3158 // `ARRAY['pg_class'::regclass]` a text array, so
3159 // `oid = ANY(…)` compared bigint against text and was
3160 // refused — while the identical `oid = 'pg_class'::regclass`
3161 // worked. Same defect the IN-list gate had, one layer down.
3162 Value::RegClass(..) | Value::RegProc(..) | Value::RegType(..) => has_bigint = true,
3163 _ => has_text = true,
3164 }
3165 }
3166 let any_numlike = has_int || has_bigint || has_numeric || has_float;
3167 if has_text || !any_numlike || (has_numeric && !numeric_representable) {
3168 let out: Vec<Option<String>> = materialised
3169 .into_iter()
3170 .map(|v| match v {
3171 Value::Null => None,
3172 Value::Text(s) | Value::Json(s) => Some(s.into_owned()),
3173 other => Some(value_to_text_for_array(&other, &ctx.render_style)),
3174 })
3175 .collect();
3176 return Ok(Value::TextArray(out));
3177 }
3178 // v7.38 (read01) — PG array-element unification across the numeric
3179 // ladder: any float → double precision[]; else any numeric →
3180 // numeric[] (each element keeps its own scale, PG's behaviour);
3181 // else the integer widths. Matches `pg_typeof(ARRAY[1, 2.5])` =
3182 // numeric[] and keeps downstream `[i]` arithmetic numeric.
3183 if has_float {
3184 let out: Vec<Option<f64>> = materialised
3185 .into_iter()
3186 .map(|v| match v {
3187 Value::Null => None,
3188 Value::Float(f) => Some(f),
3189 Value::Int(n) => Some(f64::from(n)),
3190 Value::SmallInt(n) => Some(f64::from(n)),
3191 #[allow(clippy::cast_precision_loss)]
3192 Value::BigInt(n) => Some(n as f64),
3193 #[allow(clippy::cast_precision_loss)]
3194 Value::Numeric { scaled, scale, .. } => {
3195 Some(scaled as f64 / libm::pow(10.0, f64::from(scale)))
3196 }
3197 _ => None,
3198 })
3199 .collect();
3200 return Ok(Value::FloatArray(out));
3201 }
3202 if has_numeric {
3203 let out: Vec<Option<(i128, u16)>> = materialised
3204 .into_iter()
3205 .map(|v| match v {
3206 Value::Null => None,
3207 Value::SmallInt(n) => Some((i128::from(n), 0)),
3208 Value::Int(n) => Some((i128::from(n), 0)),
3209 Value::BigInt(n) => Some((i128::from(n), 0)),
3210 Value::Numeric { scaled, scale, .. } => Some((scaled, scale)),
3211 _ => None,
3212 })
3213 .collect();
3214 return Ok(Value::NumericArray(out));
3215 }
3216 if has_bigint {
3217 let out: Vec<Option<i64>> = materialised
3218 .into_iter()
3219 .map(|v| match v {
3220 Value::Null => None,
3221 Value::Int(n) => Some(i64::from(n)),
3222 Value::SmallInt(n) => Some(i64::from(n)),
3223 Value::BigInt(n) => Some(n),
3224 // Keep in step with the `has_bigint` classification above:
3225 // whatever is counted there has to be convertible here, and
3226 // the arm below panics rather than errors. Round 652 added
3227 // the reg family to the classifier and this materialiser
3228 // took a wire-visible panic until it learned them too.
3229 Value::RegClass(oid, _) | Value::RegProc(oid, _) | Value::RegType(oid, _) => {
3230 Some(oid)
3231 }
3232 _ => unreachable!(),
3233 })
3234 .collect();
3235 return Ok(Value::BigIntArray(out));
3236 }
3237 let out: Vec<Option<i32>> = materialised
3238 .into_iter()
3239 .map(|v| match v {
3240 Value::Null => None,
3241 Value::Int(n) => Some(n),
3242 Value::SmallInt(n) => Some(i32::from(n)),
3243 _ => unreachable!(),
3244 })
3245 .collect();
3246 Ok(Value::IntArray(out))
3247}
3248
3249/// Out-of-lined `eval_expr` arm — keeps the recursive frame small
3250/// (stack-depth guard budget); body unchanged.
3251#[inline(never)]
3252fn eval_function_call_arm(
3253 name: &str,
3254 args: &[Expr],
3255 row: &Row<'static>,
3256 ctx: &EvalContext<'_>,
3257) -> Result<Value<'static>, EvalError> {
3258 // v7.39 (read01 round 77) — named arguments (`f(x := 1)` / `f(x => 1)`).
3259 // The parser leaves them in the tree because only the catalog knows a user
3260 // function's parameter names; here they become positional, once, for
3261 // builtins and user functions alike.
3262 // v7.39 (read01 round 100) — `VARIADIC <array>` splices the array's
3263 // elements in as individual trailing arguments before dispatch, so a
3264 // variadic builtin (concat / concat_ws / format / …) sees them exactly as
3265 // if they had been written out. Done before the named-arg pass and the
3266 // positional dispatch.
3267 if args.iter().any(|a| matches!(a, Expr::Variadic(_))) {
3268 let expanded = expand_variadic_args(args, row, ctx)?;
3269 return eval_function_call_arm(name, &expanded, row, ctx);
3270 }
3271 // v7.39 (round 276) — `date_part('timezone'|…, <timestamp>)` must be
3272 // REJECTED the way EXTRACT already rejects it, and the judgement
3273 // needs the argument's STATIC declared type: SPG stores timestamptz
3274 // in the same `Value::Timestamp`, so a timestamptz legitimately
3275 // answers 0 while a plain timestamp must error. The dispatch below
3276 // receives values, not expressions, so the check belongs here —
3277 // the same place, and the same r237 trust rule (only a cast or a
3278 // column is believed), that the EXTRACT arm uses.
3279 if name.eq_ignore_ascii_case("date_part")
3280 && args.len() == 2
3281 && let Expr::Literal(spg_sql::ast::Literal::String(unit)) = &args[0]
3282 && matches!(
3283 unit.to_ascii_lowercase().as_str(),
3284 "timezone" | "timezone_hour" | "timezone_minute"
3285 )
3286 && matches!(&args[1], Expr::Cast { .. } | Expr::Column(_))
3287 && let Some(sch) = crate::describe::describe_expr(&args[1], ctx.columns)
3288 && matches!(sch.ty, spg_storage::DataType::Timestamp)
3289 {
3290 return Err(EvalError::TypeMismatch {
3291 detail: alloc::format!(
3292 "unit \"{}\" not supported for type timestamp without time zone",
3293 unit.to_ascii_lowercase()
3294 ),
3295 });
3296 }
3297 // v7.39 (round 258) — `pg_typeof` over an ENUM. An enum value travels
3298 // as `Value::Text` (its label), so the value-driven namer answered
3299 // `text`; the type lives in the EXPRESSION, which this arm still has.
3300 // `expr_enum_type_name` resolves a column or a cast statically — the
3301 // same static-only discipline round 253 used for EXTRACT's type name.
3302 if name.eq_ignore_ascii_case("pg_typeof")
3303 && let [arg] = args
3304 {
3305 // The name must be a REAL enum in the catalog. `expr_enum_type_name`
3306 // returns any named cast's target verbatim — it is only a
3307 // pre-filter for `expr_enum_labels`, which does the catalog
3308 // lookup — so using it alone hijacked every `x::float8` /
3309 // `x::int2` and reported SPG's internal spelling instead of PG's.
3310 // v7.39 (round 259) — domains report their own name too; like an
3311 // enum, a domain value travels as its BASE type's value, so the
3312 // name has to come from the expression. Both lookups are gated on
3313 // the catalog: `expr_enum_type_name` returns ANY named cast's
3314 // target verbatim, so an ungated use hijacks `x::float8`.
3315 let is_user_type = |e: &Expr| {
3316 expr_enum_type_name(e, ctx.columns)
3317 .filter(|n| {
3318 // v7.39 (round 330, V48) — the information_schema
3319 // domains are built into the server rather than
3320 // catalog objects (a catalog domain is user data and
3321 // would be dumped), so they are recognised here too.
3322 crate::system_catalog::is_information_schema_domain(n)
3323 || ctx.catalog.is_some_and(|cat| {
3324 cat.enum_types().contains_key(*n)
3325 || cat.domain_types().contains_key(*n)
3326 // v7.39 (round 263) — composites too: a cast
3327 // to one reported the generic `record`.
3328 || cat.composite_types().contains_key(*n)
3329 })
3330 })
3331 .map(alloc::string::String::from)
3332 };
3333 let is_enum = is_user_type;
3334 if let Some(en) = is_enum(arg) {
3335 return Ok(Value::text(en));
3336 }
3337 // `ARRAY[<enum>, …]` reports the array form.
3338 if let Expr::Array(items) = arg
3339 && let Some(first) = items.first()
3340 && let Some(en) = is_enum(first)
3341 {
3342 return Ok(Value::text(alloc::format!("{en}[]")));
3343 }
3344 }
3345 if args.iter().any(|a| matches!(a, Expr::NamedArg { .. })) {
3346 let positional = resolve_named_args(name, args, ctx)?;
3347 return eval_function_call_arm(name, &positional, row, ctx);
3348 }
3349 eval_function_call_positional(name, args, row, ctx)
3350}
3351
3352/// v7.39 (read01 round 100) — rewrite a call's argument list, replacing each
3353/// `VARIADIC <array>` with the array's elements as literal arguments. A NULL
3354/// array contributes no elements (PG treats `VARIADIC NULL` as empty). Regular
3355/// arguments are carried through untouched so they still evaluate against the
3356/// row in the recursive call.
3357fn expand_variadic_args(
3358 args: &[Expr],
3359 row: &Row<'static>,
3360 ctx: &EvalContext<'_>,
3361) -> Result<alloc::vec::Vec<Expr>, EvalError> {
3362 let mut out = alloc::vec::Vec::with_capacity(args.len());
3363 for a in args {
3364 if let Expr::Variadic(inner) = a {
3365 let v = eval_expr(inner, row, ctx)?;
3366 let elems = crate::select::array_value_to_elements(&v).map_err(|_| {
3367 EvalError::TypeMismatch {
3368 detail: "VARIADIC argument must be an array".into(),
3369 }
3370 })?;
3371 for e in elems {
3372 out.push(Expr::Literal(crate::value_to_literal(e)));
3373 }
3374 } else {
3375 out.push(a.clone());
3376 }
3377 }
3378 Ok(out)
3379}
3380
3381/// The declared parameter names of `fname`, or `None` when it takes none.
3382/// Builtins whose parameters PG names live in the table; everything else asks
3383/// the catalog, where a user function's `args_repr` has carried its parameter
3384/// names since the day CREATE FUNCTION stored them.
3385fn declared_param_names(fname: &str, ctx: &EvalContext<'_>) -> Option<alloc::vec::Vec<String>> {
3386 let lower = fname.to_ascii_lowercase();
3387 let builtin: &[&str] = match lower.as_str() {
3388 "make_date" => &["year", "month", "day"],
3389 "make_time" => &["hour", "min", "sec"],
3390 "make_timestamp" | "make_timestamptz" => &["year", "month", "mday", "hour", "min", "sec"],
3391 "make_interval" => &["years", "months", "weeks", "days", "hours", "mins", "secs"],
3392 _ => &[],
3393 };
3394 if !builtin.is_empty() {
3395 return Some(builtin.iter().map(|s| (*s).to_string()).collect());
3396 }
3397 let cat = ctx.catalog?;
3398 let def = cat
3399 .functions()
3400 .values()
3401 .find(|f| f.name.eq_ignore_ascii_case(&lower))?;
3402 let names = spg_storage::function_arg_names(&def.args_repr);
3403 if names.iter().all(alloc::string::String::is_empty) {
3404 return None;
3405 }
3406 Some(names)
3407}
3408
3409/// Rewrite a call's arguments into positional order. Positional arguments fill
3410/// slots left to right; a named one goes to its declared slot. Slots nobody
3411/// filled stay absent for a user function (arity is checked at the call) and
3412/// become integer 0 for the `make_*` builtins, whose trailing fields PG
3413/// defaults that way.
3414fn resolve_named_args(
3415 fname: &str,
3416 args: &[Expr],
3417 ctx: &EvalContext<'_>,
3418) -> Result<alloc::vec::Vec<Expr>, EvalError> {
3419 let Some(params) = declared_param_names(fname, ctx) else {
3420 return Err(EvalError::TypeMismatch {
3421 detail: alloc::format!("function {fname}(...) does not support named arguments"),
3422 });
3423 };
3424 let mut slots: alloc::vec::Vec<Option<Expr>> = (0..params.len()).map(|_| None).collect();
3425 let mut next_positional = 0usize;
3426 for a in args {
3427 let (idx, val) = match a {
3428 Expr::NamedArg { name, expr } => {
3429 let i = params
3430 .iter()
3431 .position(|p| p.eq_ignore_ascii_case(name))
3432 .ok_or_else(|| EvalError::TypeMismatch {
3433 detail: alloc::format!("{fname}(...) has no argument named \"{name}\""),
3434 })?;
3435 (i, (**expr).clone())
3436 }
3437 other => {
3438 let i = next_positional;
3439 next_positional += 1;
3440 (i, other.clone())
3441 }
3442 };
3443 if idx >= slots.len() {
3444 return Err(EvalError::TypeMismatch {
3445 detail: alloc::format!("{fname}(...) got too many arguments"),
3446 });
3447 }
3448 if slots[idx].is_some() {
3449 return Err(EvalError::TypeMismatch {
3450 detail: alloc::format!("{fname}(...) got multiple values for one argument"),
3451 });
3452 }
3453 slots[idx] = Some(val);
3454 }
3455 let make_family = fname.to_ascii_lowercase().starts_with("make_");
3456 let mut out = alloc::vec::Vec::with_capacity(slots.len());
3457 for slot in slots {
3458 match slot {
3459 Some(e) => out.push(e),
3460 None if make_family => {
3461 out.push(Expr::Literal(spg_sql::ast::Literal::Integer(0)));
3462 }
3463 // A user function's unfilled slot is simply not passed; the call's
3464 // own arity check phrases the error.
3465 None => {}
3466 }
3467 }
3468 Ok(out)
3469}
3470
3471fn eval_function_call_positional(
3472 name: &str,
3473 args: &[Expr],
3474 row: &Row<'static>,
3475 ctx: &EvalContext<'_>,
3476) -> Result<Value<'static>, EvalError> {
3477 // v7.39 (round 237) — COALESCE / GREATEST / LEAST resolve their
3478 // arguments to one type the way CASE and ARRAY do. Checked statically:
3479 // an argument may have side effects (`COALESCE(nextval('s'), 1)`), so
3480 // its declared type is read rather than its value.
3481 if matches!(args.len(), 2..) {
3482 let construct = if name.eq_ignore_ascii_case("coalesce") {
3483 Some("COALESCE")
3484 } else if name.eq_ignore_ascii_case("greatest") {
3485 Some("GREATEST")
3486 } else if name.eq_ignore_ascii_case("least") {
3487 Some("LEAST")
3488 } else {
3489 None
3490 };
3491 if let Some(construct) = construct {
3492 unify_branch_types_static(construct, args.iter(), ctx)?;
3493 }
3494 }
3495 // v7.39 (read01 utils/adt, enum.c) — the enum introspection
3496 // v7.39 (read01 utils/adt, enum.c) — the enum introspection
3497 // family needs the ARGUMENT'S STATIC TYPE (the value is
3498 // usually NULL::enumtype): first/last/range over the
3499 // catalog's member order. Out-of-line so eval_expr's
3500 // recursion frame stays small (stack-depth guard budget).
3501 if (name.eq_ignore_ascii_case("enum_first")
3502 || name.eq_ignore_ascii_case("enum_last")
3503 || name.eq_ignore_ascii_case("enum_range"))
3504 && enum_introspection_applies(args, ctx)
3505 {
3506 return eval_enum_introspection(name, args, row, ctx);
3507 }
3508 // v7.39 (tz epic) — AT TIME ZONE (fn form: timezone(zone, ts))
3509 // with a NAMED zone needs the host tzdb and the argument's
3510 // static type for its two directions:
3511 // naive AT ZONE -> that zone's wall clock -> UTC instant
3512 // tstz AT ZONE -> UTC instant -> that zone's wall clock
3513 // Fixed offsets / abbreviations keep the legacy path below.
3514 // v7.39 (round 523) — `to_char(tstz, fmt)` renders the LOCAL clock
3515 // in the session zone, and its zone tokens name that zone. It was
3516 // rendering the UTC reading and spelling it `UTC`, so a formatted
3517 // stamp disagreed with the same value's own `::text`.
3518 if args.len() == 2
3519 && name.eq_ignore_ascii_case("to_char")
3520 && let Some(zone) = ctx.session_gucs.and_then(|g| g.get("timezone"))
3521 && !zone.eq_ignore_ascii_case("utc")
3522 && !zone.eq_ignore_ascii_case("gmt")
3523 && crate::describe::describe_expr(&args[0], ctx.columns)
3524 .is_some_and(|s| matches!(s.ty, spg_storage::DataType::Timestamptz))
3525 && let Value::Timestamp(t) = eval_expr(&args[0], row, ctx)?
3526 {
3527 let off = ctx.session_tz_offset_at(t);
3528 let abbrev = ctx
3529 .session_tz_abbrev_at(t)
3530 .unwrap_or_else(|| zone.to_uppercase());
3531 let vals = [
3532 Value::Timestamp(t.saturating_add(off)),
3533 eval_expr(&args[1], row, ctx)?,
3534 ];
3535 return crate::eval::strings::to_char_in_zone(&vals, Some((&abbrev, off)));
3536 }
3537 // v7.39 (round 523) — `date_trunc(unit, tstz)` truncates on the
3538 // LOCAL calendar in the session zone. It was truncating in UTC and
3539 // rendering the result in the session zone, so under `SET TimeZone =
3540 // 'Asia/Tokyo'` a day truncation answered `2020-01-01 09:00:00+09` —
3541 // not a day boundary at all, and the wrong day for anything before
3542 // 09:00. Every report grouped by day was cut nine hours late.
3543 //
3544 // The three-argument form already does exactly this, DST reverse
3545 // lookup and all, so the session zone is passed to THAT rather than
3546 // written a second time. Only a statically-known timestamptz shifts:
3547 // a naive timestamp has no zone to be read in.
3548 if args.len() == 2
3549 && (name.eq_ignore_ascii_case("date_trunc") || name.eq_ignore_ascii_case("date_bin"))
3550 && let Some(zone) = ctx.session_gucs.and_then(|g| g.get("timezone"))
3551 && !zone.eq_ignore_ascii_case("utc")
3552 && !zone.eq_ignore_ascii_case("gmt")
3553 && crate::describe::describe_expr(&args[1], ctx.columns)
3554 .is_some_and(|s| matches!(s.ty, spg_storage::DataType::Timestamptz))
3555 {
3556 let vals = [
3557 eval_expr(&args[0], row, ctx)?,
3558 eval_expr(&args[1], row, ctx)?,
3559 Value::text(zone.clone()),
3560 ];
3561 return datetime::date_trunc(&vals, ctx);
3562 }
3563 if args.len() == 2
3564 && name.eq_ignore_ascii_case("timezone")
3565 && let zone_v = eval_expr(&args[0], row, ctx)?
3566 && let Value::Text(zone) = &zone_v
3567 && datetime::resolve_zone_offset(zone.as_ref()).is_none()
3568 && !zone.trim().eq_ignore_ascii_case("utc")
3569 && !zone.trim().eq_ignore_ascii_case("gmt")
3570 && zone.parse::<i64>().is_err()
3571 && ctx.tz_offset_fn.is_some()
3572 {
3573 let zone = zone.trim();
3574 let src_is_tstz = crate::describe::describe_expr(&args[1], ctx.columns)
3575 .is_some_and(|s| matches!(s.ty, spg_storage::DataType::Timestamptz));
3576 let inner = eval_expr(&args[1], row, ctx)?;
3577 if let Value::Timestamp(t) = inner {
3578 if src_is_tstz {
3579 if let Some(off) = ctx.zone_offset_at(zone, t) {
3580 return Ok(Value::Timestamp(t + off));
3581 }
3582 } else if let Some(utc) = ctx.zone_local_to_utc(zone, t) {
3583 return Ok(Value::Timestamp(utc));
3584 }
3585 return Err(EvalError::TypeMismatch {
3586 detail: alloc::format!("time zone \"{zone}\" not recognized"),
3587 });
3588 }
3589 }
3590 // v7.29 (round-22 phase 3) - prefix fast path: LEFT(col, n)
3591 // on a TEXT column borrows the cell and clones only the
3592 // prefix. The generic path clones the WHOLE cell first -
3593 // a LEFT(body, 120) over 24k x 30 KB rows spent 383 ms
3594 // copying bytes it then threw away (7 ms without LEFT).
3595 if args.len() == 2
3596 && name.eq_ignore_ascii_case("left")
3597 && let Expr::Column(c) = &args[0]
3598 && let Some(cell) = resolve_column_borrowed(c, row, ctx)?
3599 {
3600 {
3601 match cell {
3602 Value::Null => return Ok(Value::Null),
3603 Value::Text(t) => {
3604 let n_v = eval_expr(&args[1], row, ctx)?;
3605 if let Value::SmallInt(_) | Value::Int(_) | Value::BigInt(_) = n_v {
3606 let n = match n_v {
3607 Value::SmallInt(x) => i64::from(x),
3608 Value::Int(x) => i64::from(x),
3609 Value::BigInt(x) => x,
3610 _ => 0,
3611 };
3612 return Ok(Value::text(text_prefix_chars(t, n)));
3613 }
3614 }
3615 _ => {}
3616 }
3617 }
3618 }
3619 // v7.38 (T-tstz Phase 1) — the ONE case where pg_typeof needs the
3620 // static type: timestamptz. The runtime value is a tz-less
3621 // Value::Timestamp, so the value-driven answer below can only ever
3622 // say "without time zone". For every other type the value-driven
3623 // path is strictly better (it distinguishes json vs jsonb, keeps
3624 // NULL as "unknown", and is not fooled by describe_expr's lossy
3625 // heuristics), so we consult the static typer ONLY when it says
3626 // Timestamptz and otherwise fall through untouched.
3627 if args.len() == 1
3628 && name.eq_ignore_ascii_case("pg_typeof")
3629 && crate::describe::describe_expr(&args[0], ctx.columns)
3630 .is_some_and(|s| matches!(s.ty, spg_storage::DataType::Timestamptz))
3631 {
3632 return Ok(Value::text::<alloc::string::String>(
3633 "timestamp with time zone".into(),
3634 ));
3635 }
3636 // v7.39 (round 694) — `oid[]`. Its VALUE is a BigIntArray, so the
3637 // value-driven namer answers `bigint[]`; the declared type lives in the
3638 // expression, exactly as it does for the Timestamptz arm above and for
3639 // the enum / domain / composite arms further down. The scalar `oid`
3640 // needed the same treatment in round 667.
3641 if args.len() == 1
3642 && name.eq_ignore_ascii_case("pg_typeof")
3643 && crate::describe::describe_expr(&args[0], ctx.columns)
3644 .is_some_and(|s| matches!(s.ty, spg_storage::DataType::OidArray))
3645 {
3646 return Ok(Value::text::<alloc::string::String>("oid[]".into()));
3647 }
3648 // v7.39 (read01 round 56) — a COMPOSITE column reports its type NAME, not
3649 // the generic `record` the runtime value would give. Composite-ness lives
3650 // outside the DataType lattice (the stored form is JSON), so the witness is
3651 // the column's `user_composite_type` — the same shape as the enum witness.
3652 if args.len() == 1
3653 && name.eq_ignore_ascii_case("pg_typeof")
3654 && let Expr::Column(c) = &args[0]
3655 && let Some(cname) = ctx
3656 .columns
3657 .iter()
3658 .find(|sc| sc.name.eq_ignore_ascii_case(&c.name))
3659 .and_then(|sc| sc.user_composite_type.as_deref())
3660 {
3661 return Ok(Value::text::<alloc::string::String>(cname.into()));
3662 }
3663 // v7.39 (round 291) — `name` is a DECLARED type over a Value::Text,
3664 // so the value can never witness it; the schema is the only witness,
3665 // exactly as for a composite column above.
3666 if args.len() == 1
3667 && name.eq_ignore_ascii_case("pg_typeof")
3668 && let Expr::Column(c) = &args[0]
3669 && ctx
3670 .columns
3671 .iter()
3672 .any(|sc| sc.name.eq_ignore_ascii_case(&c.name) && sc.ty == spg_storage::DataType::Name)
3673 {
3674 return Ok(Value::text::<alloc::string::String>("name".into()));
3675 }
3676 // …and the same for a bare `'abc'::name`, where the cast TARGET is
3677 // the witness. PG computes pg_typeof statically; SPG reads the
3678 // value, which by then is an ordinary text.
3679 if args.len() == 1
3680 && name.eq_ignore_ascii_case("pg_typeof")
3681 && let Expr::Cast {
3682 target: spg_sql::ast::CastTarget::Named(n),
3683 ..
3684 } = &args[0]
3685 && n.eq_ignore_ascii_case("name")
3686 {
3687 return Ok(Value::text::<alloc::string::String>("name".into()));
3688 }
3689 // v7.39 (read01 round 116) — a bare, uncoerced string literal is PG's
3690 // `unknown` type, not text: `pg_typeof('x')` / `pg_typeof('123')` /
3691 // `pg_typeof('2024-01-01')` all report `unknown`. The literal only becomes
3692 // text once context coerces it — a cast (`'x'::text`), a concatenation, or
3693 // a function argument — each of which is a different Expr node that falls
3694 // through to the value-driven path below (which correctly says text).
3695 if args.len() == 1
3696 && name.eq_ignore_ascii_case("pg_typeof")
3697 && matches!(&args[0], Expr::Literal(spg_sql::ast::Literal::String(_)))
3698 {
3699 return Ok(Value::text::<alloc::string::String>("unknown".into()));
3700 }
3701 // v7.37.16 — pg_typeof of a NULL cell reports the COLUMN's
3702 // static type when it has one (PG: `VALUES (NULL),(1.5)`
3703 // types the column numeric and its NULL row's pg_typeof says
3704 // numeric, not unknown).
3705 //
3706 // A BARE `NULL` stays "unknown", as PG has it. That used to fall
3707 // out of TEXT being absent from the name table — a NULL literal
3708 // describes as TEXT, so the lookup returned None and the caller
3709 // reported unknown. Round 871 filled that table in, which silently
3710 // took the bare-NULL behaviour with it and broke
3711 // `pg_typeof_null_returns_unknown`. The rule is now stated rather
3712 // than emergent: an untyped NULL literal is unknown, a NULL that
3713 // was cast reports what it was cast to.
3714 if args.len() == 1 && name.eq_ignore_ascii_case("pg_typeof") {
3715 if matches!(&args[0], Expr::Literal(spg_sql::ast::Literal::Null)) {
3716 return Ok(Value::text::<alloc::string::String>("unknown".into()));
3717 }
3718 let v = eval_expr(&args[0], row, ctx)?;
3719 if matches!(v, Value::Null)
3720 && let Some(shape) = crate::describe::describe_expr(&args[0], ctx.columns)
3721 && let Some(n) = pg_typeof_name_for_datatype(shape.ty)
3722 {
3723 return Ok(Value::text(n));
3724 }
3725 // v7.39 (round 640) — a NON-null cell normally answers from the
3726 // value, which is right for every type whose identity the value
3727 // carries. `xid8` has no value of its own — a cell is a
3728 // `Value::BigInt` — so it can only ever say "bigint" unless the
3729 // schema is asked. `xid` is listed with it because a cell that
3730 // reached here as a plain integer (a synthesised catalog row
3731 // that has not been converted) should still name its column's
3732 // type rather than the storage it arrived in.
3733 //
3734 // v7.39 (round 667) — `oid` joins them for the same reason and no
3735 // other: its cell is a `Value::BigInt` too, so `pg_typeof(1::oid)`
3736 // answered `bigint`. Still not a general switch — where the value
3737 // knows its own identity it is the better witness, because an
3738 // expression's static shape is an approximation and its result is
3739 // the fact.
3740 if !matches!(v, Value::Null)
3741 && let Some(shape) = crate::describe::describe_expr(&args[0], ctx.columns)
3742 && matches!(
3743 shape.ty,
3744 spg_storage::DataType::Xid
3745 | spg_storage::DataType::Xid8
3746 | spg_storage::DataType::Oid
3747 )
3748 && let Some(n) = pg_typeof_name_for_datatype(shape.ty)
3749 {
3750 return Ok(Value::text(n));
3751 }
3752 return apply_function(name, &[v], ctx);
3753 }
3754 // v7.37 D.1 — COALESCE result-type coercion. PG gives COALESCE the
3755 // common type of its branches, so a typed sibling (`NULL::time`,
3756 // `col::time`) makes the whole expression that type and an untyped
3757 // string-literal branch is coerced to it. Without this,
3758 // `COALESCE(NULL::time, '12:00')::text` rendered the raw `12:00`
3759 // instead of `12:00:00`. Only kicks in when the picked value is a
3760 // bare Text and a non-text cast-target sibling exists.
3761 if name.eq_ignore_ascii_case("coalesce") && !args.is_empty() {
3762 // v7.39 (round 609) — PG's COALESCE does not evaluate a branch past
3763 // the first non-NULL one. This evaluated every branch into a `Vec`
3764 // and so RAISED errors PG never raises: `coalesce(1, 1/0)`,
3765 // `coalesce(NULL, 2, 1/0)` and `coalesce(1, NULL, 1/0)` all failed
3766 // with "division by zero" where PG answers 1, 2 and 1.
3767 //
3768 // A branch after the pick is still READ for its type — that is what
3769 // decides the result's, and `COALESCE(1, 2.5)` is numeric in both
3770 // engines — but its error is discarded, because PG never runs it and
3771 // so never reports it. A branch that fails contributes no type,
3772 // which is the same as SPG having no declared type to widen to.
3773 //
3774 // The two `Vec`s this replaces cost two allocations a row even for
3775 // `coalesce(id, 0)` over a plain INTEGER column, where the answer
3776 // needs none.
3777 let mut result: Option<Value<'static>> = None;
3778 let mut tbuf = [spg_storage::DataType::Int; 8];
3779 let mut ntypes = 0usize;
3780 let mut spill: Vec<spg_storage::DataType> = Vec::new();
3781 for a in args {
3782 let v = if result.is_none() {
3783 eval_expr(a, row, ctx)?
3784 } else {
3785 match eval_expr(a, row, ctx) {
3786 Ok(v) => v,
3787 Err(_) => continue,
3788 }
3789 };
3790 // v7.39 (round 649) — a NULL branch still has a TYPE, and PG
3791 // resolves COALESCE's result from the branches' declared
3792 // types, not from the values that survive. `Value::Null` has
3793 // no `data_type()`, so `coalesce(1::int, NULL::float8)`
3794 // collected only `integer` and answered integer where PG
3795 // answers double precision. Ask the expression when the
3796 // value cannot say — inside the arm that already runs, and
3797 // only on the NULL that would otherwise contribute nothing.
3798 let branch_ty = match v.data_type() {
3799 Some(t) => Some(t),
3800 None => crate::describe::describe_expr(a, ctx.columns).map(|sh| sh.ty),
3801 };
3802 if let Some(t) = branch_ty {
3803 if ntypes < tbuf.len() {
3804 tbuf[ntypes] = t;
3805 ntypes += 1;
3806 } else {
3807 spill.push(t);
3808 }
3809 }
3810 if result.is_none() && !matches!(v, Value::Null) {
3811 result = Some(v);
3812 }
3813 }
3814 let result = result.unwrap_or(Value::Null);
3815 if matches!(result, Value::Text(_)) {
3816 if let Some(target) = args.iter().find_map(coalesce_type_hint) {
3817 return crate::eval::cast::cast_value(result, target);
3818 }
3819 }
3820 // v7.38 (read01) — otherwise widen the picked value to the PG
3821 // common type of all branches (COALESCE(1, 2.5) → numeric).
3822 if spill.is_empty() {
3823 return Ok(widen_to_common(result, &tbuf[..ntypes]));
3824 }
3825 let mut types: Vec<spg_storage::DataType> = tbuf[..ntypes].to_vec();
3826 types.append(&mut spill);
3827 return Ok(widen_to_common(result, &types));
3828 }
3829 let evaluated: Result<Vec<Value<'static>>, _> =
3830 args.iter().map(|a| eval_expr(a, row, ctx)).collect();
3831 let evaluated = evaluated?;
3832 // v7.39 (read01 json.c) — to_json(timestamptz) spells the instant in
3833 // ISO 8601 WITH the session-zone offset ("2024-03-09T14:05:06+00:00"),
3834 // unlike plain timestamp. The runtime value carries no tz tag, so the
3835 // argument's static type is the witness.
3836 if (name.eq_ignore_ascii_case("to_json") || name.eq_ignore_ascii_case("to_jsonb"))
3837 && evaluated.len() == 1
3838 && let Some(Value::Timestamp(t)) = evaluated.first()
3839 && args.first().is_some_and(|a| {
3840 crate::describe::describe_expr(a, ctx.columns)
3841 .is_some_and(|sh| matches!(sh.ty, spg_storage::DataType::Timestamptz))
3842 })
3843 {
3844 let off = ctx.session_tz_offset_at(*t);
3845 let local = t + off;
3846 let days = local.div_euclid(86_400_000_000);
3847 let day_us = local.rem_euclid(86_400_000_000);
3848 let (y, mo, d) = civil_from_days(i32::try_from(days).unwrap_or(0));
3849 let secs = day_us / 1_000_000;
3850 let frac = day_us % 1_000_000;
3851 let (hh, mi, ss) = (secs / 3600, (secs / 60) % 60, secs % 60);
3852 let mut txt = alloc::format!("{y:04}-{mo:02}-{d:02}T{hh:02}:{mi:02}:{ss:02}");
3853 if frac != 0 {
3854 let f = alloc::format!("{frac:06}");
3855 txt.push('.');
3856 txt.push_str(f.trim_end_matches('0'));
3857 }
3858 let (sign, omag) = if off < 0 { ('-', -off) } else { ('+', off) };
3859 let (oh, om) = (omag / 3_600_000_000, (omag / 60_000_000) % 60);
3860 let _ = core::fmt::Write::write_fmt(&mut txt, format_args!("{sign}{oh:02}:{om:02}"));
3861 return Ok(Value::json(alloc::format!("\"{txt}\"")));
3862 }
3863 // v7.39 (enum order knife) — greatest/least over enum-typed arguments
3864 // pick by member order, not label text (PG). The witness needs the arg
3865 // ASTs, so this can't live in the value-level function dispatch.
3866 if (name.eq_ignore_ascii_case("greatest") || name.eq_ignore_ascii_case("least"))
3867 && let Some(labels) = args
3868 .iter()
3869 .find_map(|a| expr_enum_labels(a, ctx.columns, ctx.catalog))
3870 && evaluated
3871 .iter()
3872 .all(|v| matches!(v, Value::Text(_) | Value::Null))
3873 {
3874 let is_greatest = name.eq_ignore_ascii_case("greatest");
3875 let mut best: Option<&Value<'static>> = None;
3876 for v in evaluated.iter().filter(|v| !matches!(v, Value::Null)) {
3877 best = Some(match best {
3878 None => v,
3879 Some(b) => match enum_ord_cmp(labels, v, b) {
3880 Some(core::cmp::Ordering::Greater) if is_greatest => v,
3881 Some(core::cmp::Ordering::Less) if !is_greatest => v,
3882 Some(_) => b,
3883 // A non-member snuck in — fall out to the generic path.
3884 None => return apply_function(name, &evaluated, ctx),
3885 },
3886 });
3887 }
3888 return Ok(best.cloned().unwrap_or(Value::Null));
3889 }
3890 // v7.39 (round 693) — and the same for a declared COLLATION, which
3891 // `least`/`greatest` need for the same structural reason: the witness
3892 // is the argument's column, so it cannot live in the value-level
3893 // dispatch either. Measured on PG18 over a column declaring
3894 // en_US.utf8: `least(a,'d')` is `d` and `greatest(a,'d')` is `Zebra`,
3895 // where byte order gives the pair reversed.
3896 if (name.eq_ignore_ascii_case("greatest") || name.eq_ignore_ascii_case("least"))
3897 && evaluated
3898 .iter()
3899 .all(|v| matches!(v, Value::Text(_) | Value::Null))
3900 && let Some(coll) = greatest_least_collation(args, ctx)
3901 {
3902 let is_greatest = name.eq_ignore_ascii_case("greatest");
3903 let mut best: Option<&Value<'static>> = None;
3904 for v in evaluated.iter().filter(|v| !matches!(v, Value::Null)) {
3905 best = Some(match (best, v) {
3906 (None, _) => v,
3907 (Some(Value::Text(y)), Value::Text(x)) => {
3908 match crate::collate::compare(&coll, x, y) {
3909 Some(core::cmp::Ordering::Greater) if is_greatest => v,
3910 Some(core::cmp::Ordering::Less) if !is_greatest => v,
3911 Some(_) => best.unwrap_or(v),
3912 // Not a collation this build performs after all —
3913 // one answer, from the generic path.
3914 None => return apply_function(name, &evaluated, ctx),
3915 }
3916 }
3917 (Some(b), _) => b,
3918 });
3919 }
3920 return Ok(best.cloned().unwrap_or(Value::Null));
3921 }
3922 // v7.39 (round 621) — an unadorned string literal takes the type the
3923 // function's parameter asks for. `justify_interval('36 hours')` is answered
3924 // by PG and was refused here, and so were `justify_days('35 days')` and
3925 // `justify_hours('27 hours')` — the spelling everyone writes, since typing
3926 // `INTERVAL` in front of the literal is exactly what PG saves you from.
3927 //
3928 // Only a LITERAL is resolved, which is the same boundary round 620 drew for
3929 // the boolean connectives: `justify_interval(t)` over a TEXT column stays
3930 // refused, because PG refuses it too (no such overload). The arg ASTs are
3931 // needed to tell those apart, so this cannot live in the value-level
3932 // dispatch — the same reason the enum witness above sits here.
3933 if let Some(want) = unknown_literal_param_type(name) {
3934 let mut coerced = evaluated;
3935 for (i, a) in args.iter().enumerate() {
3936 if is_unknown_string_literal(a)
3937 && let Some(slot) = coerced.get_mut(i)
3938 {
3939 *slot = cast::cast_value_in(
3940 core::mem::replace(slot, Value::Null),
3941 want.clone(),
3942 false,
3943 )?;
3944 }
3945 }
3946 return apply_function(name, &coerced, ctx);
3947 }
3948 apply_function(name, &evaluated, ctx)
3949}
3950
3951/// v7.39 (round 621) — the parameter type a bare string literal resolves to.
3952///
3953/// PG resolves an `unknown` literal to whatever the chosen overload declares.
3954/// SPG has no overload resolution to hang that on, so the functions whose only
3955/// parameter is unambiguous are listed. `None` leaves the argument alone.
3956fn unknown_literal_param_type(name: &str) -> Option<spg_sql::ast::CastTarget> {
3957 match name.to_ascii_lowercase().as_str() {
3958 "justify_days" | "justify_hours" | "justify_interval" => {
3959 Some(spg_sql::ast::CastTarget::Interval)
3960 }
3961 _ => None,
3962 }
3963}
3964
3965/// Out-of-lined `eval_expr` arm — keeps the recursive frame small
3966/// (stack-depth guard budget); body unchanged.
3967#[inline(never)]
3968fn eval_any_all_arm(
3969 expr: &Expr,
3970 op: &BinOp,
3971 array: &Expr,
3972 is_any: bool,
3973 row: &Row<'static>,
3974 ctx: &EvalContext<'_>,
3975) -> Result<Value<'static>, EvalError> {
3976 let lhs = eval_expr(expr, row, ctx)?;
3977 let arr = eval_expr(array, row, ctx)?;
3978 any_all_over(lhs, arr, op, is_any)
3979}
3980
3981/// v7.39 (round 597) — the ANY/ALL comparison with both sides already in
3982/// hand. Split out so a CONSTANT right-hand array can be built once at
3983/// compile time instead of once per row: `WHERE id = ANY (ARRAY[1..10])`
3984/// rebuilt the array for all 500k rows and cost 268 ms against PG18's 8.3,
3985/// rising to 494 ms at twenty elements, while the equivalent
3986/// `id IN (1..10)` took 2.3. The body below is unchanged; it never touched
3987/// `row`.
3988pub(crate) fn any_all_over(
3989 lhs: Value<'static>,
3990 arr: Value<'static>,
3991 op: &BinOp,
3992 is_any: bool,
3993) -> Result<Value<'static>, EvalError> {
3994 if matches!(arr, Value::Null) {
3995 return Ok(Value::Null);
3996 }
3997 // v7.38 (read01) — an unknown-string RHS (`x = ANY('{1,2,3}')`)
3998 // takes the LHS's type: coerce the external array text to the array
3999 // type matching the LHS's element type, like PG.
4000 let arr = match &arr {
4001 Value::Text(_) => {
4002 // The LHS's element type, or TEXT when the LHS is an
4003 // untyped NULL (PG's unknown → text default).
4004 let arr_ty = match lhs.data_type() {
4005 Some(spg_storage::DataType::SmallInt) => spg_storage::DataType::SmallIntArray,
4006 Some(spg_storage::DataType::Int) => spg_storage::DataType::IntArray,
4007 Some(spg_storage::DataType::BigInt) => spg_storage::DataType::BigIntArray,
4008 Some(spg_storage::DataType::Numeric { .. }) => spg_storage::DataType::NumericArray,
4009 Some(spg_storage::DataType::Float) => spg_storage::DataType::FloatArray,
4010 Some(spg_storage::DataType::Bool) => spg_storage::DataType::BoolArray,
4011 Some(spg_storage::DataType::Date) => spg_storage::DataType::DateArray,
4012 _ => spg_storage::DataType::TextArray,
4013 };
4014 crate::conversions::coerce_value(arr.clone(), arr_ty, "", 0).unwrap_or(arr)
4015 }
4016 _ => arr,
4017 };
4018 // Build the element list generically so every scalar array type
4019 // (numeric[], float8[], bool[], date[], …) is accepted, not just
4020 // int/bigint/text.
4021 let Some(len) = array_len(&arr) else {
4022 return Err(EvalError::TypeMismatch {
4023 detail: format!(
4024 "ANY/ALL right-hand side must be an array, got {}",
4025 crate::conversions::pg_type_name_for_error_opt(arr.data_type())
4026 ),
4027 });
4028 };
4029 let elems: Vec<Option<Value>> = (0..len)
4030 .map(|i| match array_element_at(&arr, i) {
4031 Some(Value::Null) | None => None,
4032 Some(v) => Some(v),
4033 })
4034 .collect();
4035 // PG: `x op ANY (empty)` → false and `x op ALL (empty)` →
4036 // true, decided purely by emptiness — the comparison is
4037 // never evaluated, so a NULL LHS is irrelevant. This must
4038 // short-circuit before `saw_null` is seeded from the LHS,
4039 // otherwise `NULL op ANY/ALL (empty)` wrongly yields NULL.
4040 if elems.is_empty() {
4041 return Ok(Value::Bool(!is_any));
4042 }
4043 let mut saw_null = matches!(lhs, Value::Null);
4044 let mut saw_match = false;
4045 let mut saw_mismatch = false;
4046 for elem in elems {
4047 let elem_v = match elem {
4048 Some(v) => v,
4049 None => {
4050 saw_null = true;
4051 continue;
4052 }
4053 };
4054 if matches!(lhs, Value::Null) {
4055 saw_null = true;
4056 continue;
4057 }
4058 match apply_binary(*op, lhs.clone(), elem_v) {
4059 Ok(Value::Bool(true)) => saw_match = true,
4060 Ok(Value::Bool(false)) => saw_mismatch = true,
4061 Ok(Value::Null) => saw_null = true,
4062 Ok(other) => {
4063 return Err(EvalError::TypeMismatch {
4064 detail: format!(
4065 "ANY/ALL comparison didn't return Bool: {}",
4066 crate::conversions::pg_type_name_for_error_opt(other.data_type())
4067 ),
4068 });
4069 }
4070 Err(e) => return Err(e),
4071 }
4072 }
4073 let result = if is_any {
4074 if saw_match {
4075 Value::Bool(true)
4076 } else if saw_null {
4077 Value::Null
4078 } else {
4079 Value::Bool(false)
4080 }
4081 } else if saw_mismatch {
4082 Value::Bool(false)
4083 } else if saw_null {
4084 Value::Null
4085 } else {
4086 Value::Bool(true)
4087 };
4088 Ok(result)
4089}
4090
4091/// Out-of-lined `eval_expr` arm — keeps the recursive frame small
4092/// (stack-depth guard budget); body unchanged.
4093#[inline(never)]
4094fn eval_case_arm(
4095 operand: &Option<alloc::boxed::Box<Expr>>,
4096 branches: &[(Expr, Expr)],
4097 else_branch: &Option<alloc::boxed::Box<Expr>>,
4098 row: &Row<'static>,
4099 ctx: &EvalContext<'_>,
4100) -> Result<Value<'static>, EvalError> {
4101 // v7.39 (round 237) — PG resolves the RESULT branches to one type and
4102 // refuses the CASE when they have no common one, before running
4103 // anything. SPG returned whichever branch fired, so
4104 // `CASE WHEN true THEN 1 ELSE 'a'::text END` answered `1` and the same
4105 // expression answered text on another row.
4106 {
4107 // PG resolves the ELSE branch FIRST and then the WHEN results, and
4108 // its message names the running type before the conflicting one —
4109 // which is why `THEN 1 ELSE 'a'::text` reports "text and integer"
4110 // while a two-WHEN `THEN 1 ... THEN true` reports "integer and
4111 // boolean". Probed against 18.4; the order is observable.
4112 let mut results: Vec<&Expr> = Vec::with_capacity(branches.len() + 1);
4113 if let Some(e) = else_branch {
4114 results.push(e);
4115 }
4116 results.extend(branches.iter().map(|(_, r)| r));
4117 unify_branch_types_static("CASE", results, ctx)?;
4118 }
4119 let operand_value = match operand {
4120 Some(o) => Some(eval_expr(o, row, ctx)?),
4121 None => None,
4122 };
4123 // v7.37 D.1 — CASE result-type coercion (same rule as COALESCE): a
4124 // typed result branch (`... THEN '10:00'::time`) makes the whole
4125 // CASE that type, so an untyped string-literal branch is coerced to
4126 // it. Compute the hint once from every THEN/ELSE branch.
4127 let case_hint = branches
4128 .iter()
4129 .map(|(_, t)| t)
4130 .chain(else_branch.iter().map(|b| b.as_ref()))
4131 .find_map(coalesce_type_hint);
4132 // v7.38 (read01) — the CASE result is PG's common type of every
4133 // THEN/ELSE branch, so a taken integer branch is widened to
4134 // numeric when a sibling branch is numeric (and `pg_typeof` /
4135 // downstream division match PG). Only one branch is evaluated, so
4136 // the type must come from the branch expressions statically.
4137 let branch_types: Vec<spg_storage::DataType> = branches
4138 .iter()
4139 .map(|(_, t)| t)
4140 .chain(else_branch.iter().map(|b| b.as_ref()))
4141 .filter_map(|e| crate::describe::describe_expr(e, ctx.columns).map(|s| s.ty))
4142 .collect();
4143 let coerce = |v: Value<'static>| -> Result<Value<'static>, EvalError> {
4144 let v = match (&v, &case_hint) {
4145 (Value::Text(_), Some(target)) => cast::cast_value(v, target.clone())?,
4146 _ => v,
4147 };
4148 Ok(widen_to_common(v, &branch_types))
4149 };
4150 for (when_expr, then_expr) in branches {
4151 let when_value = eval_expr(when_expr, row, ctx)?;
4152 let matched = match &operand_value {
4153 // v7.39 (round 346, M1) — the WHEN condition is a truth value,
4154 // not a boolean-shaped one: `CASE WHEN 1 THEN 'a' END` used to
4155 // answer NULL in BOTH dialects, where MariaDB answers `a` and
4156 // PG raises `argument of CASE/WHEN must be type boolean`.
4157 None => predicate_is_true(&when_value, "CASE/WHEN", ctx.mysql_dialect)?,
4158 // v7.39 (round 412) — under the MySQL default collation the
4159 // `CASE op WHEN v` equality folds Text/BpChar operands (CI +
4160 // accent + PAD SPACE), matching `op = v` outside CASE.
4161 Some(op_v) => {
4162 let (l, r) = if ctx.mysql_dialect {
4163 match (op_v, &when_value) {
4164 (Value::Text(x), Value::Text(y)) | (Value::BpChar(x), Value::BpChar(y)) => {
4165 (
4166 Value::text(spg_storage::mysql_compare_fold(x)),
4167 Value::text(spg_storage::mysql_compare_fold(y)),
4168 )
4169 }
4170 _ => (op_v.clone(), when_value),
4171 }
4172 } else {
4173 (op_v.clone(), when_value)
4174 };
4175 matches!(
4176 apply_binary(spg_sql::ast::BinOp::Eq, l, r)?,
4177 Value::Bool(true)
4178 )
4179 }
4180 };
4181 if matched {
4182 return coerce(eval_expr(then_expr, row, ctx)?);
4183 }
4184 }
4185 match else_branch {
4186 Some(e) => coerce(eval_expr(e, row, ctx)?),
4187 None => Ok(Value::Null),
4188 }
4189}
4190
4191/// Out-of-lined `eval_expr` arm — keeps the recursive frame small
4192/// (stack-depth guard budget); body unchanged.
4193#[inline(never)]
4194fn eval_array_slice_arm(
4195 target: &Expr,
4196 lo: &Option<alloc::boxed::Box<Expr>>,
4197 hi: &Option<alloc::boxed::Box<Expr>>,
4198 row: &Row<'static>,
4199 ctx: &EvalContext<'_>,
4200) -> Result<Value<'static>, EvalError> {
4201 let target_v = eval_expr(target, row, ctx)?;
4202 if matches!(target_v, Value::Null) {
4203 return Ok(Value::Null);
4204 }
4205 let bound = |e: Option<&Expr>| -> Result<Option<i64>, EvalError> {
4206 match e {
4207 None => Ok(None),
4208 Some(b) => match eval_expr(b, row, ctx)? {
4209 Value::Null => Ok(None),
4210 Value::Int(n) => Ok(Some(i64::from(n))),
4211 Value::BigInt(n) => Ok(Some(n)),
4212 Value::SmallInt(n) => Ok(Some(i64::from(n))),
4213 other => Err(EvalError::TypeMismatch {
4214 detail: format!(
4215 "array slice bound must be integer, got {}",
4216 crate::conversions::pg_type_name_for_error_opt(other.data_type())
4217 ),
4218 }),
4219 },
4220 }
4221 };
4222 let lo_b = bound(lo.as_deref())?;
4223 let hi_b = bound(hi.as_deref())?;
4224 fn window(len: usize, lo: Option<i64>, hi: Option<i64>) -> (usize, usize) {
4225 let start = lo.map_or(0, |l| (l.max(1) - 1) as usize).min(len);
4226 let end = hi.map_or(len, |h| h.max(0) as usize).min(len);
4227 (start, end.max(start))
4228 }
4229 match target_v {
4230 Value::TextArray(items) => {
4231 let (s, e) = window(items.len(), lo_b, hi_b);
4232 Ok(Value::TextArray(items[s..e].to_vec()))
4233 }
4234 Value::IntArray(items) => {
4235 let (s, e) = window(items.len(), lo_b, hi_b);
4236 Ok(Value::IntArray(items[s..e].to_vec()))
4237 }
4238 Value::BigIntArray(items) => {
4239 let (s, e) = window(items.len(), lo_b, hi_b);
4240 Ok(Value::BigIntArray(items[s..e].to_vec()))
4241 }
4242 other => Err(EvalError::TypeMismatch {
4243 detail: format!(
4244 "slice target must be an array, got {}",
4245 crate::conversions::pg_type_name_for_error_opt(other.data_type())
4246 ),
4247 }),
4248 }
4249}
4250
4251/// Out-of-lined `eval_expr` arm — keeps the recursive frame small
4252/// (stack-depth guard budget); body unchanged.
4253#[inline(never)]
4254fn eval_in_list_arm(
4255 expr: &Expr,
4256 list: &[Expr],
4257 negated: bool,
4258 row: &Row<'static>,
4259 ctx: &EvalContext<'_>,
4260) -> Result<Value<'static>, EvalError> {
4261 // v7.39 (round 238) — PG resolves the whole list's type BEFORE comparing
4262 // anything, so `1 IN (1, 'a'::text)` is refused. SPG compared item by
4263 // item and broke on the first match, so the offending element was never
4264 // reached and the predicate quietly answered true. Checked statically,
4265 // like round 237: evaluating the rest of the list to inspect it would
4266 // change when side effects fire.
4267 require_in_list_comparable(expr, list, ctx)?;
4268 // v7.39 (round 364, M4 P2) — a MySQL session folds text before the
4269 // membership test, so `t IN ('FOO')` matches 'Foo'. `BINARY` is not
4270 // reachable through a bare column needle here; the fold is text-only.
4271 // v7.39 (round 370, M4 P4a) — an explicit `COLLATE utf8mb4_bin` needle
4272 // column is byte-wise, so it does not fold. v7.39 (round 371, M4 P4b) —
4273 // a per-expression `… COLLATE utf8mb4_bin` / `BINARY …` on the needle
4274 // OR any list item forces the whole membership test byte-wise.
4275 let in_fold = ctx.mysql_dialect
4276 && !resolve::operand_is_binary_column(expr, ctx)
4277 && !resolve::is_binary_coerced(expr)
4278 && !list.iter().any(|i| resolve::is_binary_coerced(i));
4279 let needle = mysql_collation_key(eval_expr(expr, row, ctx)?, in_fold);
4280 let needle_null = matches!(needle, Value::Null);
4281 let mut saw_null = needle_null && !list.is_empty();
4282 let mut matched = false;
4283 if !needle_null {
4284 for item in list {
4285 let v = mysql_collation_key(eval_expr(item, row, ctx)?, in_fold);
4286 if matches!(v, Value::Null) {
4287 saw_null = true;
4288 continue;
4289 }
4290 match apply_binary(BinOp::Eq, needle.clone(), v)? {
4291 Value::Bool(true) => {
4292 matched = true;
4293 break;
4294 }
4295 Value::Bool(false) => {}
4296 Value::Null => saw_null = true,
4297 other => {
4298 return Err(EvalError::TypeMismatch {
4299 detail: format!(
4300 "IN comparison didn't return Bool: {}",
4301 crate::conversions::pg_type_name_for_error_opt(other.data_type())
4302 ),
4303 });
4304 }
4305 }
4306 }
4307 }
4308 let inner = if matched {
4309 Value::Bool(true)
4310 } else if saw_null {
4311 Value::Null
4312 } else {
4313 Value::Bool(false)
4314 };
4315 Ok(match (negated, inner) {
4316 (true, Value::Bool(b)) => Value::Bool(!b),
4317 (_, v) => v,
4318 })
4319}
4320
4321/// Out-of-lined `eval_expr` arm — keeps the recursive frame small
4322/// (stack-depth guard budget); body unchanged.
4323#[inline(never)]
4324fn eval_like_arm(
4325 expr: &Expr,
4326 pattern: &Expr,
4327 negated: bool,
4328 case_insensitive: bool,
4329 row: &Row<'static>,
4330 ctx: &EvalContext<'_>,
4331) -> Result<Value<'static>, EvalError> {
4332 let v = eval_expr(expr, row, ctx)?;
4333 let p = eval_expr(pattern, row, ctx)?;
4334 // NULL on either side propagates to NULL — same as PG.
4335 // v7.39 (bpchar epic) — LIKE matches bpchar on its PADDED
4336 // stored form, per PG's bpchar pattern operators.
4337 let (text, pat) = match (v, p) {
4338 (Value::Null, _) | (_, Value::Null) => return Ok(Value::Null),
4339 (Value::Text(a) | Value::BpChar(a), Value::Text(b) | Value::BpChar(b)) => (a, b),
4340 (Value::Text(_) | Value::BpChar(_), other) | (other, _) => {
4341 return Err(EvalError::TypeMismatch {
4342 detail: format!(
4343 "LIKE requires text operands, got {}",
4344 crate::conversions::pg_type_name_for_error_opt(other.data_type())
4345 ),
4346 });
4347 }
4348 };
4349 // v7.25 (round-17) — ILIKE folds both operands (PG
4350 // lowercases per the default collation).
4351 // v7.39 (round 364, M4 P2) — a MySQL session's default collation is
4352 // accent- and case-insensitive, so `LIKE` folds both sides the way
4353 // `ILIKE` does; the wildcards `%` / `_` are not Latin letters so the
4354 // fold leaves them alone.
4355 // v7.39 (round 370, M4 P4a) — an explicit `COLLATE utf8mb4_bin` column
4356 // matches byte-wise, so it does not fold. v7.39 (round 371, M4 P4b) —
4357 // a per-expression `… COLLATE utf8mb4_bin` / `BINARY …` on either the
4358 // value or the pattern forces byte-wise too.
4359 let mysql = ctx.mysql_dialect
4360 && !resolve::operand_is_binary_column(expr, ctx)
4361 && !resolve::operand_is_binary_column(pattern, ctx)
4362 && !resolve::is_binary_coerced(expr)
4363 && !resolve::is_binary_coerced(pattern);
4364 let m = if case_insensitive {
4365 like_match(&text.to_lowercase(), &pat.to_lowercase())?
4366 } else if mysql {
4367 like_match(
4368 &spg_storage::mysql_ci_fold(&text),
4369 &spg_storage::mysql_ci_fold(&pat),
4370 )?
4371 } else {
4372 like_match(&text, &pat)?
4373 };
4374 Ok(Value::Bool(if negated { !m } else { m }))
4375}
4376
4377/// Out-of-lined `eval_expr` arm — keeps the recursive frame small
4378/// (stack-depth guard budget); body unchanged.
4379#[inline(never)]
4380fn eval_extract_arm(
4381 field: &spg_sql::ast::ExtractField,
4382 source: &Expr,
4383 row: &Row<'static>,
4384 ctx: &EvalContext<'_>,
4385) -> Result<Value<'static>, EvalError> {
4386 let v = eval_expr(source, row, ctx)?;
4387 extract_from_value(field, v, source, ctx)
4388}
4389
4390/// v7.39 (round 595) — the field extraction, with the source value already
4391/// in hand. Split out so the compiled-predicate program can pop the source
4392/// off its stack instead of handing the whole node back to the interpreter:
4393/// one non-compilable node used to disqualify the entire WHERE, and
4394/// `WHERE extract(year FROM t) = 2020` was interpreting the column read and
4395/// the comparison too. The body below is unchanged; it never touched `row`.
4396pub(crate) fn extract_from_value(
4397 field: &spg_sql::ast::ExtractField,
4398 v: Value<'static>,
4399 source: &Expr,
4400 ctx: &EvalContext<'_>,
4401) -> Result<Value<'static>, EvalError> {
4402 // v7.39 (round 382) — MySQL coerces a date/time STRING to its temporal
4403 // value for EXTRACT (`EXTRACT(YEAR FROM '2020-05-15')` is 2020, and the
4404 // time fields read a `'... HH:MM:SS'` string); PG needs a typed source.
4405 let v = match &v {
4406 Value::Text(s) if ctx.mysql_dialect => text_as_temporal(s).unwrap_or(v),
4407 _ => v,
4408 };
4409 // v7.39 (tz epic) — timezone[_hour|_minute] of a timestamptz
4410 // reports the SESSION offset at that instant (PG: 32400 for
4411 // Tokyo; -14400 for New York in July).
4412 if matches!(
4413 field,
4414 spg_sql::ast::ExtractField::Timezone
4415 | spg_sql::ast::ExtractField::TimezoneHour
4416 | spg_sql::ast::ExtractField::TimezoneMinute
4417 ) && let Value::Timestamp(t) = &v
4418 && crate::describe::describe_expr(source, ctx.columns)
4419 .is_some_and(|s| matches!(s.ty, spg_storage::DataType::Timestamptz))
4420 {
4421 let off_secs = ctx.session_tz_offset_at(*t) / 1_000_000;
4422 let n = match field {
4423 spg_sql::ast::ExtractField::Timezone => off_secs,
4424 spg_sql::ast::ExtractField::TimezoneHour => off_secs / 3600,
4425 _ => (off_secs / 60) % 60,
4426 };
4427 // v7.39 (round 253) — numeric, like every other EXTRACT result.
4428 return Ok(Value::Numeric {
4429 scaled: i128::from(n),
4430 scale: 0,
4431 kind: spg_storage::NumericKind::Finite,
4432 });
4433 }
4434 // v7.39 (round 523) — and every OTHER field of a timestamptz reads
4435 // the local clock in the session zone, which is the whole reason PG
4436 // has the type. `extract(hour from …)` answered the UTC hour under
4437 // `SET TimeZone = 'Asia/Tokyo'` — 0 where PG says 9 — and
4438 // `extract(dow …)` therefore named the wrong DAY, so a report
4439 // grouped by weekday put nine hours of every Sunday under Saturday.
4440 // Only fields of the local clock shift; epoch and julian are
4441 // absolute, and the timezone fields answered above.
4442 let v = match &v {
4443 Value::Timestamp(t)
4444 if !matches!(
4445 field,
4446 spg_sql::ast::ExtractField::Epoch
4447 | spg_sql::ast::ExtractField::Julian
4448 | spg_sql::ast::ExtractField::Timezone
4449 | spg_sql::ast::ExtractField::TimezoneHour
4450 | spg_sql::ast::ExtractField::TimezoneMinute
4451 ) && crate::describe::describe_expr(source, ctx.columns)
4452 .is_some_and(|s| matches!(s.ty, spg_storage::DataType::Timestamptz)) =>
4453 {
4454 Value::Timestamp(t.saturating_add(ctx.session_tz_offset_at(*t)))
4455 }
4456 _ => v,
4457 };
4458 // v7.39 (round 253) — the source's PG type name for error wording,
4459 // upgraded from the declared type when statically known (a tstz
4460 // VALUE is indistinguishable from a timestamp). Only a cast /
4461 // column is trusted (the r237 lesson: describe_expr reports a
4462 // binary operator as its left operand's type).
4463 let static_declared = matches!(source, Expr::Cast { .. } | Expr::Column(_))
4464 .then(|| crate::describe::describe_expr(source, ctx.columns))
4465 .flatten()
4466 .map(|sch| sch.ty);
4467 let src_name = match static_declared {
4468 Some(spg_storage::DataType::Timestamptz) => "timestamp with time zone",
4469 _ => datetime::value_src_type_name(&v),
4470 };
4471 // PG rejects the timezone family on a plain timestamp (0A000);
4472 // only reject when the declared type is STATICALLY timestamp — a
4473 // dynamic value stays lenient (the pre-r253 zero answer).
4474 if matches!(
4475 field,
4476 spg_sql::ast::ExtractField::Timezone
4477 | spg_sql::ast::ExtractField::TimezoneHour
4478 | spg_sql::ast::ExtractField::TimezoneMinute
4479 ) && matches!(static_declared, Some(spg_storage::DataType::Timestamp))
4480 {
4481 return Err(EvalError::TypeMismatch {
4482 detail: alloc::format!(
4483 "unit \"{}\" not supported for type timestamp without time zone",
4484 alloc::format!("{field}").to_lowercase()
4485 ),
4486 });
4487 }
4488 // v7.39 (round 418) — MySQL's compound units (`DAY_SECOND`, `YEAR_MONTH`,
4489 // …) reach here as `ExtractField::Other`, which PG rejects. Under the
4490 // MySQL dialect they pack several components into one integer instead.
4491 if ctx.mysql_dialect
4492 && let spg_sql::ast::ExtractField::Other(name) = field
4493 && let Some(packed) = crate::eval::datetime::mysql_compound_extract(name, &v)
4494 {
4495 return Ok(packed);
4496 }
4497 extract_field(field, &v, src_name)
4498}
4499
4500/// Out-of-lined `eval_expr` arm — keeps the recursive frame small
4501/// (stack-depth guard budget); body unchanged.
4502#[inline(never)]
4503fn eval_array_subscript_arm(
4504 expr: &Expr,
4505 row: &Row<'static>,
4506 ctx: &EvalContext<'_>,
4507) -> Result<Value<'static>, EvalError> {
4508 // Collect the whole subscript chain so PG's multi-dimensional
4509 // access (`arr[i][j]` is ONE N-subscript op) is distinguishable
4510 // from chained 1-D indexing. `arr[1][2]` parses as
4511 // `(arr[1])[2]`; PG indexes the matrix directly and returns NULL
4512 // for a partial subscript (`arr[1]` on a 2-D array is NULL).
4513 let mut idx_exprs: Vec<&Expr> = Vec::new();
4514 let mut base = expr;
4515 while let Expr::ArraySubscript { target, index } = base {
4516 idx_exprs.push(index);
4517 base = target;
4518 }
4519 idx_exprs.reverse();
4520 let base_v = eval_expr(base, row, ctx)?;
4521 if matches!(
4522 base_v,
4523 Value::IntArray2D(_)
4524 | Value::BigIntArray2D(_)
4525 | Value::TextArray2D(_)
4526 | Value::BoolArray2D(_)
4527 ) {
4528 return eval_matrix_subscript(&base_v, &idx_exprs, row, ctx);
4529 }
4530 // 1-D array / JSON: apply each subscript left-to-right. This
4531 // reproduces the prior single-subscript semantics exactly, and
4532 // chained JSON (`j['a']['b']`) still resolves step by step.
4533 let mut cur = base_v;
4534 for ix in idx_exprs {
4535 cur = apply_one_subscript(cur, ix, row, ctx)?;
4536 }
4537 Ok(cur)
4538}
4539
4540/// Out-of-lined `eval_expr` arm — keeps the recursive frame small
4541/// (stack-depth guard budget); body unchanged.
4542#[inline(never)]
4543fn eval_field_access_arm(
4544 base: &Expr,
4545 field: &str,
4546 row: &Row<'static>,
4547 ctx: &EvalContext<'_>,
4548) -> Result<Value<'static>, EvalError> {
4549 // v7.38 (read01, T9) — composite field access `(expr).field`.
4550 // The base evaluates to a record; look the member up by name
4551 // (`f1`..`fN` for an anonymous ROW, base column names for a
4552 // whole-row). A NULL record yields NULL (PG semantics).
4553 let v = eval_expr(base, row, ctx)?;
4554 match v {
4555 Value::Null => Ok(Value::Null),
4556 Value::Composite(fields) => fields
4557 .into_iter()
4558 .find(|(name, _)| name == field)
4559 .map(|(_, val)| val)
4560 .ok_or_else(|| missing_field_error(base, field, ctx)),
4561 _ => Err(not_a_composite_error(base, field, ctx)),
4562 }
4563}
4564
4565/// v7.39 (round 285) — PG words a missing composite field three ways, and
4566/// which one you get depends on the base expression's STATIC type, not on
4567/// the value:
4568///
4569/// * a named composite — `column "nosuch" not found in data type rc9`
4570/// * a whole-row table reference — `column rt8.nosuch does not exist`
4571/// (unquoted, and qualified — the odd one out)
4572/// * an anonymous ROW or `::record` — `could not identify column
4573/// "nosuch" in record data type`
4574///
4575/// All three read off live PG 18.4. A `Value::Composite` carries its field
4576/// names but not its type name, so the base expression is what decides.
4577fn missing_field_error(base: &Expr, field: &str, ctx: &EvalContext<'_>) -> EvalError {
4578 // v7.39 (round 307, V25) — the named type may arrive by cast OR from
4579 // the schema of a column that a projection produced, so ask once and
4580 // let the catalog say whether it is a composite. Before this only
4581 // the cast spelling was recognised, which is why a composite that
4582 // came through a derived table or a CTE — where the base is a plain
4583 // column — fell through to the anonymous-record wording.
4584 if let Some(name) = base_named_type(base, ctx)
4585 && ctx
4586 .catalog
4587 .is_some_and(|c| c.composite_types().contains_key(name))
4588 {
4589 return EvalError::TypeMismatch {
4590 detail: alloc::format!("column \"{field}\" not found in data type {name}"),
4591 };
4592 }
4593 if let Expr::Column(c) = base
4594 && c.qualifier.is_none()
4595 {
4596 // A column DECLARED as a named composite reports that type — the
4597 // schema records it in `user_composite_type`, which is the only
4598 // place the name survives (a `Value::Composite` does not carry it).
4599 if let Some(name) = ctx
4600 .columns
4601 .iter()
4602 .find(|col| col.name.eq_ignore_ascii_case(&c.name))
4603 .and_then(|col| col.user_composite_type.as_ref())
4604 {
4605 return EvalError::TypeMismatch {
4606 detail: alloc::format!("column \"{field}\" not found in data type {name}"),
4607 };
4608 }
4609 // A whole-row reference to a real table is the odd wording out:
4610 // qualified, and unquoted.
4611 if ctx.catalog.is_some_and(|cat| cat.get(&c.name).is_some()) {
4612 return EvalError::TypeMismatch {
4613 detail: alloc::format!("column {}.{field} does not exist", c.name),
4614 };
4615 }
4616 }
4617 EvalError::TypeMismatch {
4618 detail: alloc::format!("could not identify column \"{field}\" in record data type"),
4619 }
4620}
4621
4622/// v7.39 (round 307, V25) — the user-declared type name behind a field
4623/// access, if any: either written as a cast (`ROW(…)::rc9`) or carried on
4624/// the column's schema.
4625///
4626/// All three schema slots are consulted rather than just the composite
4627/// one. A projection currently files the name of a `::rc9` cast under
4628/// `user_enum_type` — `expr_enum_type_name` answers for ANY named cast,
4629/// without asking whether the name is an enum — so keying off one slot
4630/// would answer for some shapes and not others. The caller decides what
4631/// the name MEANS by asking the catalog, which is the only thing that
4632/// actually knows; this function just finds it.
4633fn base_named_type<'c>(base: &'c Expr, ctx: &'c EvalContext<'_>) -> Option<&'c str> {
4634 match base {
4635 Expr::Cast {
4636 target: CastTarget::Named(name),
4637 ..
4638 } => Some(name.as_str()),
4639 Expr::Column(c) => ctx
4640 .columns
4641 .iter()
4642 .find(|sc| sc.name.eq_ignore_ascii_case(&c.name))
4643 .and_then(|sc| {
4644 sc.user_composite_type
4645 .as_deref()
4646 .or(sc.user_domain_type.as_deref())
4647 .or(sc.user_enum_type.as_deref())
4648 }),
4649 _ => None,
4650 }
4651}
4652
4653/// v7.39 (round 307, V25) — PG's wording when field notation is applied
4654/// to something that is not a composite at all. It names the type:
4655/// `column notation .f applied to type pos9, which is not a composite
4656/// type` — for a domain and an enum alike. Only when the base has no
4657/// user-declared type at all does the generic message stand.
4658fn not_a_composite_error(base: &Expr, field: &str, ctx: &EvalContext<'_>) -> EvalError {
4659 if let Some(name) = base_named_type(base, ctx)
4660 && ctx.catalog.is_some_and(|c| {
4661 c.enum_types().contains_key(name) || c.domain_types().contains_key(name)
4662 })
4663 {
4664 return EvalError::TypeMismatch {
4665 detail: alloc::format!(
4666 "column notation .{field} applied to type {name}, which is not a composite type"
4667 ),
4668 };
4669 }
4670 EvalError::TypeMismatch {
4671 detail: alloc::format!("field access `.{field}` requires a composite (record) value"),
4672 }
4673}
4674
4675/// Out-of-lined `eval_expr` arm — keeps the recursive frame small
4676/// (stack-depth guard budget); body unchanged.
4677#[inline(never)]
4678/// v7.39 (round 328, V45) — the three-valued boolean tests. None of them
4679/// ever answers NULL: a NULL input is "not true" and "not false", and IS
4680/// UNKNOWN is precisely the NULL case. Verified against PG 18.4 —
4681/// `NULL::bool IS TRUE` is false, `IS NOT TRUE` true, `IS UNKNOWN` true,
4682/// and `false IS NOT FALSE` false.
4683fn eval_bool_test_arm(
4684 expr: &Expr,
4685 value: Option<bool>,
4686 negated: bool,
4687 row: &Row<'static>,
4688 ctx: &EvalContext<'_>,
4689) -> Result<Value<'static>, EvalError> {
4690 let v = eval_expr(expr, row, ctx)?;
4691 let hit = match (value, &v) {
4692 // IS UNKNOWN — the input is NULL.
4693 (None, Value::Null) => true,
4694 // v7.39 (round 625) — and PG rejects a non-boolean here too:
4695 // `argument of IS UNKNOWN must be type boolean`. MySQL has no
4696 // IS UNKNOWN, so there is no dialect branch.
4697 (None, Value::Bool(_)) => false,
4698 (None, other) => {
4699 return Err(EvalError::TypeMismatch {
4700 detail: alloc::format!(
4701 "argument of IS {}UNKNOWN must be type boolean, not type {}",
4702 if negated { "NOT " } else { "" },
4703 crate::conversions::pg_type_name_for_error_opt(other.data_type())
4704 ),
4705 });
4706 }
4707 (Some(_), Value::Null) => false,
4708 (Some(want), Value::Bool(b)) => *b == want,
4709 // v7.39 (round 397) — MySQL reads a non-boolean as a truth value
4710 // for `IS TRUE` / `IS FALSE` (`5 IS TRUE` is 1, `0 IS FALSE` is 1,
4711 // `'abc' IS TRUE` is 0). PG rejects a non-boolean at parse time, so
4712 // this only fires under the dialect; a NULL is already handled.
4713 (Some(want), other) if ctx.mysql_dialect => mysql_truthy(other) == want,
4714 // v7.39 (round 625, S05b/F29) — on PG a non-boolean is REJECTED, and
4715 // the comment above said so while the arm below answered `false`
4716 // anyway. `1 IS TRUE` came back false, which reads as "the test was
4717 // run and did not hold" rather than "you cannot ask this of an
4718 // integer" — the wrong answer for every non-boolean type, eight of
4719 // them measured. PG's own sentence, which names the operator and the
4720 // type it got.
4721 (Some(_), other) => {
4722 return Err(EvalError::TypeMismatch {
4723 detail: alloc::format!(
4724 "argument of IS {}{} must be type boolean, not type {}",
4725 if negated { "NOT " } else { "" },
4726 if value == Some(true) { "TRUE" } else { "FALSE" },
4727 crate::conversions::pg_type_name_for_error_opt(other.data_type())
4728 ),
4729 });
4730 }
4731 };
4732 Ok(Value::Bool(hit != negated))
4733}
4734
4735fn eval_is_null_arm(
4736 expr: &Expr,
4737 negated: bool,
4738 row: &Row<'static>,
4739 ctx: &EvalContext<'_>,
4740) -> Result<Value<'static>, EvalError> {
4741 // v7.38 (read01 P4.11) — `ROW(...) IS [NOT] NULL` is evaluated
4742 // field-wise, not as a whole-value null test: a row IS NULL when
4743 // every field is null, and IS NOT NULL when every field is
4744 // non-null — so the two are NOT simple negations (ROW(1,NULL) is
4745 // neither). A field that is itself a row is a non-null value, so
4746 // the check does not recurse. The `(a, b) IS NULL` tuple spelling
4747 // is already desugared to `a IS NULL AND b IS NULL` in the parser;
4748 // this covers the explicit `ROW(...)` constructor.
4749 if let Expr::FunctionCall { name, args } = expr
4750 && name.eq_ignore_ascii_case("row")
4751 {
4752 let mut all_null = true;
4753 let mut all_non_null = true;
4754 for a in args {
4755 if matches!(eval_expr(a, row, ctx)?, Value::Null) {
4756 all_non_null = false;
4757 } else {
4758 all_null = false;
4759 }
4760 }
4761 return Ok(Value::Bool(if negated { all_non_null } else { all_null }));
4762 }
4763 // v7.39 (round 962) — the same field-wise rule for a row-valued
4764 // EXPRESSION, not just the `ROW(...)` spelling. P4.11 keyed on the
4765 // syntax, so every other way to hold a row got the whole-value test:
4766 // measured against PG18.4, `SELECT an IS NULL FROM an` on a row whose
4767 // every column is NULL answered `t` there and `f` here, and a column
4768 // declared with a composite type behaved the same way. Round 961 made
4769 // whole-row references reachable through a projection, which is what
4770 // surfaced it.
4771 //
4772 // Fields are tested exactly as the `ROW(...)` arm tests its
4773 // arguments, without recursing — a field that is itself a row is a
4774 // non-null value.
4775 let v = eval_expr(expr, row, ctx)?;
4776 if let Value::Composite(fields) = &v {
4777 let mut all_null = true;
4778 let mut all_non_null = true;
4779 for (_, f) in fields {
4780 if matches!(f, Value::Null) {
4781 all_non_null = false;
4782 } else {
4783 all_null = false;
4784 }
4785 }
4786 return Ok(Value::Bool(if negated { all_non_null } else { all_null }));
4787 }
4788 let is_null = matches!(v, Value::Null);
4789 Ok(Value::Bool(if negated { !is_null } else { is_null }))
4790}
4791
4792pub fn eval_expr(
4793 expr: &Expr,
4794 row: &Row<'static>,
4795 ctx: &EvalContext<'_>,
4796) -> Result<Value<'static>, EvalError> {
4797 // v7.38 (read01 P3.25) — guard against a native stack overflow on a
4798 // pathologically nested expression (`a AND a AND … ` × thousands): the
4799 // recursion base is seeded on the outermost call, and once a deeper
4800 // call has consumed more than the budget we return an error the way
4801 // PG's check_stack_depth() does, rather than aborting the process.
4802 let sp = eval_stack_ptr();
4803 let base = ctx.recursion_base.get();
4804 if base == 0 {
4805 ctx.recursion_base.set(sp);
4806 } else if base.saturating_sub(sp) > MAX_EVAL_STACK_BYTES {
4807 return Err(EvalError::StackDepthExceeded);
4808 }
4809 match expr {
4810 Expr::AggregateOrdered { .. } => Err(EvalError::TypeMismatch {
4811 detail: "aggregate ORDER BY is only valid inside an aggregating SELECT".into(),
4812 }),
4813 // A named argument is only meaningful inside a call, where the callee's
4814 // parameter names give it a slot. Anywhere else it is a syntax error,
4815 // and saying so beats silently evaluating it as if the name were absent.
4816 Expr::NamedArg { name, .. } => Err(EvalError::TypeMismatch {
4817 detail: alloc::format!("named argument \"{name}\" is only valid in a function call"),
4818 }),
4819 Expr::Literal(l) => Ok(literal_to_value(l)),
4820 Expr::Column(c) => resolve_column(c, row, ctx),
4821 Expr::Placeholder(n) => {
4822 let idx = usize::from(*n).saturating_sub(1);
4823 ctx.params
4824 .get(idx)
4825 .cloned()
4826 .ok_or_else(|| EvalError::PlaceholderOutOfRange {
4827 n: *n,
4828 bound: u16::try_from(ctx.params.len()).unwrap_or(u16::MAX),
4829 })
4830 }
4831 // v7.39 (round 620) — an unadorned string literal carries PG's
4832 // `unknown` type, and a boolean connective is a context that resolves
4833 // it TO boolean. `'true' AND true`, `'f' OR false` and `NOT 'a'` are
4834 // answered by PG (`t`, `f`, and the input-syntax error respectively)
4835 // and were all refused here with `argument of … must be type boolean,
4836 // not type text` — the message PG reserves for an operand that really
4837 // IS text (`''::TEXT AND true`), which stays refused. Out-of-line and
4838 // behind a literal-shaped guard: this is the recursive frame the
4839 // 768 KiB stack budget is tuned against.
4840 Expr::Unary {
4841 op: spg_sql::ast::UnOp::Not,
4842 expr,
4843 } if !ctx.mysql_dialect && is_unknown_string_literal(expr) => apply_unary(
4844 spg_sql::ast::UnOp::Not,
4845 coerce_unknown_literal_to_bool(expr)?,
4846 ),
4847 Expr::Unary { op, expr } => {
4848 let v = eval_expr(expr, row, ctx)?;
4849 // The MySQL-specific unary readings (NOT any truth value, `-`/`~`
4850 // on a string, `~` unsigned) live out-of-line: `eval_expr` is the
4851 // recursive frame the 768 KiB stack-depth budget is tuned
4852 // against, and locals added here cost one nesting level each (the
4853 // round-305 / round-383 frame cliff).
4854 if ctx.mysql_dialect {
4855 if let Some(r) = mysql_unary_arm(*op, &v) {
4856 return r;
4857 }
4858 }
4859 apply_unary(*op, v)
4860 }
4861 // v7.39 (round 346, M1) — MariaDB reads both sides of AND / OR as
4862 // truth values (`1 AND 2` is 1, measured). apply_binary has no
4863 // dialect, so the coercion happens here, where it does. The body
4864 // is out-of-line: `eval_expr` is the recursive frame the 768 KiB
4865 // stack-depth budget is tuned against, and locals added here cost
4866 // one nesting level each (the round-305 frame cliff).
4867 Expr::Binary { lhs, op, rhs }
4868 if ctx.mysql_dialect && matches!(op, BinOp::And | BinOp::Or | BinOp::LogicalXor) =>
4869 {
4870 eval_mysql_connective(lhs, *op, rhs, row, ctx)
4871 }
4872 // v7.39 (round 620/621) — the unknown-literal resolution and the
4873 // short circuit, both out of line. Placed AFTER the MySQL arm so the
4874 // dialect keeps its own reading of these connectives.
4875 Expr::Binary { lhs, op, rhs } if matches!(op, BinOp::And | BinOp::Or) => {
4876 eval_connective(lhs, *op, rhs, row, ctx)
4877 }
4878 Expr::Binary { lhs, op, rhs } => {
4879 // v7.32 (P4 borrow channel) — comparison fast path. A pure
4880 // comparison op only reads its operands and returns Bool,
4881 // and for non-NUMERIC / non-INTERVAL / non-CI-collation
4882 // operands `apply_binary` IS just the NULL-3VL check plus
4883 // the ref-based `compare` (NUMERIC routes through fixed-
4884 // point `apply_binary_numeric`; INTERVAL through
4885 // `apply_binary_interval`; CI columns fold). So read the
4886 // operands borrowed — a column cell is no longer cloned
4887 // just to compare it (`WHERE thread_id != ''` alone cloned
4888 // one Text cell per scanned row). Anything that needs the
4889 // owned path falls through unchanged.
4890 if matches!(
4891 op,
4892 BinOp::Eq | BinOp::NotEq | BinOp::Lt | BinOp::LtEq | BinOp::Gt | BinOp::GtEq
4893 ) {
4894 let lc = eval_expr_cow(lhs, row, ctx)?;
4895 let rc = eval_expr_cow(rhs, row, ctx)?;
4896 // v7.39 (enum order knife) — enum-typed operands compare by
4897 // member order, not label text. Cold unless both sides are
4898 // Text and the catalog has enum types at all.
4899 if matches!(lc.as_ref(), Value::Text(_)) && matches!(rc.as_ref(), Value::Text(_)) {
4900 if let Some(r) = enum_compare_hook(*op, lhs, rhs, lc.as_ref(), rc.as_ref(), ctx)
4901 {
4902 return r;
4903 }
4904 // v7.39 (round 693) — and the collation hook, under the
4905 // same Text/Text gate for the same reason.
4906 if let Some(r) =
4907 collate_compare_hook(*op, lhs, rhs, lc.as_ref(), rc.as_ref(), ctx)
4908 {
4909 return r;
4910 }
4911 }
4912 // v7.39 (round 351, M11) — the three conditions fold into
4913 // ONE call. Adding a fourth as another `||` here tipped the
4914 // 768 KiB stack guard on its own (measured — this is the
4915 // hottest recursive frame there is); one call is cheaper
4916 // than the three it replaces.
4917 let owned_path = needs_owned_compare(lc.as_ref(), rc.as_ref(), lhs, rhs, ctx);
4918 if !owned_path {
4919 if lc.as_ref().is_null() || rc.as_ref().is_null() {
4920 return Ok(Value::Null);
4921 }
4922 return compare(*op, lc.as_ref(), rc.as_ref()).map_err(|e| {
4923 unknown_literal_cmp_error(e, lhs, rhs, lc.as_ref(), rc.as_ref())
4924 });
4925 }
4926 let (l, r) = collation_fold_for_compare(
4927 *op,
4928 lhs,
4929 rhs,
4930 lc.into_owned(),
4931 rc.into_owned(),
4932 ctx,
4933 );
4934 // The owned call consumes the values; the rewrite needs the
4935 // literal's text and the other side's type only on the ERROR
4936 // path, so capture those two up front — the capture itself is
4937 // gated on the cheap expr test, so a comparison with no
4938 // unknown literal pays one branch.
4939 let probe = (is_unknown_string_literal(lhs) || is_unknown_string_literal(rhs))
4940 .then(|| (l.clone(), r.clone()));
4941 return apply_binary_in(*op, l, r, ctx.mysql_dialect).map_err(|e| match &probe {
4942 Some((pl, pr)) => unknown_literal_cmp_error(e, lhs, rhs, pl, pr),
4943 None => e,
4944 });
4945 }
4946 let l = eval_expr(lhs, row, ctx)?;
4947 let r = eval_expr(rhs, row, ctx)?;
4948 // v7.17.0 Phase 2.5 — collation-aware text comparison.
4949 // When either operand of a comparison op references a
4950 // column declared `COLLATE "case_insensitive"` (or any
4951 // MySQL `_ci` collation), case-fold both sides before
4952 // the byte-wise compare so `WHERE name = 'foo'` matches
4953 // stored `'Foo'`. Non-Text values fall straight through
4954 // — the helper is a no-op outside Text-Text equality
4955 // and inequality.
4956 let (l, r) = collation_fold_for_compare(*op, lhs, rhs, l, r, ctx);
4957 // v7.39 (GUC knife 4) — `date/interval/float || text` textifies
4958 // through the out-functions, which honour the session render
4959 // style. Pre-render the style-sensitive operand here (the
4960 // orthodox home is an implicit-cast node at type resolution;
4961 // until then this keeps apply_binary style-free). Default
4962 // style short-circuits — text_concat's own value_to_text
4963 // produces the identical bytes.
4964 if matches!(op, spg_sql::ast::BinOp::Concat)
4965 && ctx.render_style != format::RenderStyle::default()
4966 {
4967 let styled = |v: Value<'static>| -> Value<'static> {
4968 match &v {
4969 Value::Date(_)
4970 | Value::Timestamp(_)
4971 | Value::Interval { .. }
4972 | Value::Float(_)
4973 | Value::Real(_) => {
4974 Value::text(values::value_to_text_styled(&v, &ctx.render_style))
4975 }
4976 _ => v,
4977 }
4978 };
4979 let (sl, sr) = (styled(l), styled(r));
4980 return apply_binary(*op, sl, sr);
4981 }
4982 apply_binary_mysql_unsigned(*op, lhs, rhs, l, r, ctx)
4983 }
4984 Expr::Cast { expr, target } => eval_cast_arm(expr, target, row, ctx),
4985 Expr::FieldAccess { base, field } => eval_field_access_arm(base, field, row, ctx),
4986 Expr::IsNull { expr, negated } => eval_is_null_arm(expr, *negated, row, ctx),
4987 // v7.39 (round 328, V45) — `x IS [NOT] TRUE | FALSE | UNKNOWN`.
4988 // Out-of-line like its IS NULL neighbour: an inline body here
4989 // grows every frame of the recursive evaluator, which is what
4990 // tipped the 512KB depth guard in round 305.
4991 Expr::BoolTest {
4992 expr,
4993 value,
4994 negated,
4995 } => eval_bool_test_arm(expr, *value, *negated, row, ctx),
4996 Expr::FunctionCall { name, args } => eval_function_call_arm(name, args, row, ctx),
4997 // v7.39 (read01 round 100) — VARIADIC is spliced into its enclosing
4998 // call before the args are evaluated (see eval_function_call_arm); a
4999 // bare one reaching here was written outside a function call.
5000 Expr::Variadic(_) => Err(EvalError::TypeMismatch {
5001 detail: "VARIADIC is only valid as a function-call argument".into(),
5002 }),
5003 Expr::Like {
5004 expr,
5005 pattern,
5006 negated,
5007 case_insensitive,
5008 } => eval_like_arm(expr, pattern, *negated, *case_insensitive, row, ctx),
5009 Expr::Extract { field, source } => eval_extract_arm(field, source, row, ctx),
5010 // v4.10: subquery nodes should have been resolved into
5011 // Literal / InList nodes by Engine::resolve_select_subqueries
5012 // before the row loop. Anything reaching here is a bug.
5013 Expr::ScalarSubquery(_)
5014 | Expr::Exists { .. }
5015 | Expr::InSubquery { .. }
5016 | Expr::RowInSubquery { .. }
5017 | Expr::RowCmpSubquery { .. } => Err(EvalError::TypeMismatch {
5018 detail: "subquery reached row eval — engine resolver bug".into(),
5019 }),
5020 // v7.30.2 (mailrs round-25) — flat `expr [NOT] IN (a, b, …)`.
5021 // Iterative scan with PG three-valued logic: TRUE on the first
5022 // Eq match; if nothing matched, NULL when the needle is NULL or
5023 // any comparison was NULL; FALSE otherwise. Empty list (only
5024 // reachable via an empty subquery result) is FALSE / TRUE even
5025 // for a NULL needle — no comparison ever happens.
5026 Expr::InList {
5027 expr,
5028 list,
5029 negated,
5030 } => eval_in_list_arm(expr, list, *negated, row, ctx),
5031 // v4.12: window functions should have been rewritten into
5032 // synthetic __win_N column references by
5033 // exec_select_with_window before row eval. Anything
5034 // reaching here is similarly a bug.
5035 Expr::WindowFunction { .. } => Err(EvalError::TypeMismatch {
5036 detail: "window function reached row eval — engine rewrite bug".into(),
5037 }),
5038 // v7.10.10 — `ARRAY[expr, expr, …]` constructor.
5039 // v7.11.13 — element-type detection: all integers →
5040 // IntArray (or BigIntArray when widening), any Text →
5041 // TextArray. Non-TEXT non-integer elements (Bool, Float)
5042 // stringify into TextArray as the safe default.
5043 Expr::Array(items) => eval_array_arm(items, row, ctx),
5044 // v7.10.12 — `arr[i]` PG-style 1-based indexing.
5045 // Out-of-range indices (including i ≤ 0) return NULL.
5046 Expr::ArraySubscript { .. } => eval_array_subscript_arm(expr, row, ctx),
5047 // Array slice `arr[lo:hi]` — PG 1-based, both ends
5048 // inclusive, out-of-range bounds clamp, missing bounds
5049 // extend to the array's ends. Result keeps the element
5050 // type; an empty window yields an empty array.
5051 Expr::ArraySlice { target, lo, hi } => eval_array_slice_arm(target, lo, hi, row, ctx),
5052 // v7.10.12 — `x op ANY(arr)` / `x op ALL(arr)`. PG
5053 // 3VL: ANY → true if any element compares-true; NULL if
5054 // no true but some NULL; false otherwise. ALL: false if
5055 // any compares-false; NULL if no false but some NULL;
5056 // true otherwise.
5057 Expr::AnyAll {
5058 expr,
5059 op,
5060 array,
5061 is_any,
5062 } => eval_any_all_arm(expr, op, array, *is_any, row, ctx),
5063 // v7.13.0 — CASE WHEN … END (mailrs round-5 G9).
5064 // Short-circuit on the first matching branch. Searched form
5065 // (operand=None) treats each branch's WHEN as a Bool
5066 // predicate. Simple form (operand=Some) compares with =.
5067 // ELSE on no match; NULL if no ELSE.
5068 Expr::Case {
5069 operand,
5070 branches,
5071 else_branch,
5072 } => eval_case_arm(operand, branches, else_branch, row, ctx),
5073 }
5074}
5075
5076/// v7.10.10 — best-effort text rendering for non-TEXT array
5077/// elements (numbers, bools, etc.). The PG rule is that
5078/// `ARRAY[1, 2]` is `int[]`, but SPG's v7.10 only models TEXT[],
5079/// so we widen by stringifying. NUMERIC formatting goes through
5080/// the existing canonical helpers to stay consistent with
5081/// `format_numeric` / `format_date` etc.
5082/// v7.37 D.1 — the COALESCE result-type hint: a sibling branch's explicit
5083/// cast target (`NULL::time`, `col::time`), unless it is Text (Text carries no
5084/// coercion). Returns the first non-Text `CastTarget` found, mirroring PG's
5085/// left-to-right common-type resolution for the common single-typed-branch case.
5086fn coalesce_type_hint(e: &Expr) -> Option<CastTarget> {
5087 match e {
5088 Expr::Cast { target, .. } if !matches!(target, CastTarget::Text) => Some(target.clone()),
5089 _ => None,
5090 }
5091}
5092
5093/// v7.38 (read01) — widen `v` to the already-resolved common type `common`
5094/// of a `CASE`/`COALESCE`/`GREATEST`/`LEAST`/`NULLIF` result, so the value's
5095/// type matches the one PG reports and downstream operators (e.g. `/`) see
5096/// the widened type (integer division vs numeric division). Only widens;
5097/// anything already at the common type, or that fails to coerce, is returned
5098/// untouched (this must never turn a working expression into an error).
5099/// NUMERIC is scale-preserving: an existing exact-numeric keeps its own scale
5100/// (PG renders `COALESCE(1.50, 2)` as `1.50`); only integers promote, to
5101/// scale 0.
5102pub(crate) fn widen_value_to(v: Value<'static>, common: spg_storage::DataType) -> Value<'static> {
5103 use spg_storage::DataType as DT;
5104 // Only widen numeric- and temporal-category results: these are the ones
5105 // whose type actually changes a downstream value (integer vs numeric
5106 // division, date vs timestamp). Widening a string result (varchar ∪ text)
5107 // would only relabel the type while risking a spurious length-limit error
5108 // when coercing into a modelled-length varchar/char, so leave it as-is.
5109 if !matches!(
5110 common,
5111 DT::SmallInt
5112 | DT::Int
5113 | DT::BigInt
5114 | DT::Numeric { .. }
5115 // v7.39 (round 649) — `real` was absent here too, so even once
5116 // `common_type` learned to rank it, the value was handed back
5117 // unwidened: `coalesce(1::int, 1::real)` stayed integer where
5118 // PG says real. Two lists, one ladder — the gap had to be
5119 // closed in both.
5120 | DT::Real
5121 | DT::Float
5122 | DT::Date
5123 | DT::Time
5124 | DT::Timestamp
5125 | DT::Timestamptz
5126 ) {
5127 return v;
5128 }
5129 if matches!(v, Value::Null) {
5130 return v;
5131 }
5132 if v.data_type() == Some(common) {
5133 return v;
5134 }
5135 if matches!(common, spg_storage::DataType::Numeric { .. })
5136 && matches!(v, Value::Numeric { .. } | Value::NumericBig(_))
5137 {
5138 return v;
5139 }
5140 let target = if matches!(common, spg_storage::DataType::Numeric { .. }) {
5141 spg_storage::DataType::Numeric {
5142 precision: 0,
5143 scale: 0,
5144 }
5145 } else {
5146 common
5147 };
5148 match crate::conversions::coerce_value(v.clone(), target, "", 0) {
5149 Ok(cv) => cv,
5150 Err(_) => v,
5151 }
5152}
5153
5154/// Widen `v` to the PG common type of the sibling `types`, or leave it as-is
5155/// when the types don't resolve to a single widening type. See
5156/// [`widen_value_to`] and [`crate::describe::common_type`].
5157pub(crate) fn widen_to_common(
5158 v: Value<'static>,
5159 types: &[spg_storage::DataType],
5160) -> Value<'static> {
5161 match crate::describe::common_type(types) {
5162 Some(common) => widen_value_to(v, common),
5163 None => v,
5164 }
5165}
5166
5167pub(crate) fn value_to_text_for_array(v: &Value, style: &format::RenderStyle) -> String {
5168 match v {
5169 Value::Text(s) | Value::Json(s) => s.to_string(),
5170 Value::Int(n) => n.to_string(),
5171 Value::BigInt(n) => n.to_string(),
5172 Value::SmallInt(n) => n.to_string(),
5173 // PG renders booleans in array external form as `t` / `f`
5174 // (the bool type's output function), not `true` / `false`.
5175 Value::Bool(b) => {
5176 if *b {
5177 "t".into()
5178 } else {
5179 "f".into()
5180 }
5181 }
5182 Value::Float(x) => format::format_float_styled(*x, style),
5183 Value::Real(x) => format::format_real_styled(*x, style),
5184 Value::Date(d) => format::format_date_styled(*d, style),
5185 Value::Timestamp(t) => format::format_timestamp_styled(*t, style),
5186 Value::Numeric {
5187 scaled,
5188 scale,
5189 kind,
5190 } => format_numeric_kind(*kind, *scaled, *scale),
5191 // v7.39 — everything else renders its canonical PG text (this
5192 // Debug fallback is how `ARRAY['\xff'::bytea]` printed
5193 // `{Bytes([255])}` on the wire).
5194 _ => values::value_to_text_styled(v, style),
5195 }
5196}
5197
5198/// SQL `LIKE` matcher. Wildcards are `%` (any run, possibly empty) and `_`
5199/// (exactly one char). `\` escapes the next pattern char so `\%` matches a
5200/// literal `%`. Matches the whole input — no implicit anchoring needed
5201/// since SQL `LIKE` is always full-string. Errs on a trailing unpaired
5202/// escape the matcher actually reaches with text left (PG's lazy 22025).
5203fn like_match(text: &str, pattern: &str) -> Result<bool, EvalError> {
5204 let pat: Vec<char> = pattern.chars().collect();
5205 like_match_str(text, &pat, 0)
5206}
5207
5208/// v7.37.16 — pg_typeof spelling for a STATIC column type (the
5209/// NULL-cell fallback; the value-level table is `pg_typeof_name`).
5210/// TEXT maps to None because a NULL literal describes as TEXT — the
5211/// unknown stand-in — and pg_typeof(NULL) must stay "unknown"; every
5212/// type not listed also returns None (caller keeps the value answer).
5213pub(crate) fn pg_typeof_name_for_datatype(t: spg_storage::DataType) -> Option<&'static str> {
5214 use spg_storage::DataType as D;
5215 Some(match t {
5216 D::SmallInt => "smallint",
5217 D::Int => "integer",
5218 D::BigInt => "bigint",
5219 D::Float => "double precision",
5220 D::Real => "real",
5221 D::Numeric { .. } => "numeric",
5222 D::Bool => "boolean",
5223 D::Date => "date",
5224 D::Time => "time without time zone",
5225 D::Timestamp => "timestamp without time zone",
5226 D::Timestamptz => "timestamp with time zone",
5227 D::Name => "name",
5228 D::Xid => "xid",
5229 D::Xid8 => "xid8",
5230 D::Oid => "oid",
5231 // v7.39 (round 694) — and its array, for the same reason.
5232 D::OidArray => "oid[]",
5233 D::Uuid => "uuid",
5234 D::Interval => "interval",
5235 // v7.39 (round 871) — the rest of what a NULL cast can be
5236 // annotated with. This table decided which types survived
5237 // `pg_typeof(NULL::t)`: the twenty above answered, everything
5238 // else fell to `_ => None` and reported `unknown`. That reads
5239 // as a NULL problem and is not one — `NULL::uuid` was right all
5240 // along while `NULL::text` was wrong, because one was listed
5241 // and the other was not.
5242 //
5243 // Names are PG18's own, taken from running `pg_typeof` there
5244 // rather than from memory: `bit varying` not `varbit`, `bit`
5245 // for any width, `character varying`, `"char"` quoted.
5246 D::Text => "text",
5247 D::Multirange(k) => match k {
5248 spg_storage::RangeKind::Int4 => "int4multirange",
5249 spg_storage::RangeKind::Int8 => "int8multirange",
5250 spg_storage::RangeKind::Num => "nummultirange",
5251 spg_storage::RangeKind::Ts => "tsmultirange",
5252 spg_storage::RangeKind::TsTz => "tstzmultirange",
5253 spg_storage::RangeKind::Date => "datemultirange",
5254 },
5255 D::Varchar(_) => "character varying",
5256 // PG names `char(n)` "character"; the one-byte internal type
5257 // spelled `"char"` is a DIFFERENT type there, and SPG maps both
5258 // onto `Char(u32)` — so this arm must answer for the declared
5259 // one. Round 871's first attempt said `"char"` here and would
5260 // have reported `char(5)` as the internal type.
5261 D::Char(_) => "character",
5262 D::Json => "json",
5263 D::Jsonb => "jsonb",
5264 D::Bytes => "bytea",
5265 D::Inet => "inet",
5266 D::Cidr => "cidr",
5267 D::Macaddr => "macaddr",
5268 D::Macaddr8 => "macaddr8",
5269 D::Bit(_) => "bit",
5270 D::BitVarying(_) => "bit varying",
5271 D::Xml => "xml",
5272 D::Money => "money",
5273 D::Point => "point",
5274 D::Lseg => "lseg",
5275 D::Path => "path",
5276 D::PgBox => "box",
5277 D::Polygon => "polygon",
5278 D::Line => "line",
5279 D::Circle => "circle",
5280 D::TextArray => "text[]",
5281 D::IntArray => "integer[]",
5282 D::BigIntArray => "bigint[]",
5283 D::SmallIntArray => "smallint[]",
5284 D::FloatArray => "double precision[]",
5285 D::NumericArray => "numeric[]",
5286 D::BoolArray => "boolean[]",
5287 D::DateArray => "date[]",
5288 D::TimestampArray => "timestamp without time zone[]",
5289 D::TimestamptzArray => "timestamp with time zone[]",
5290 D::UuidArray => "uuid[]",
5291 D::JsonArray => "json[]",
5292 D::JsonbArray => "jsonb[]",
5293 D::BytesArray => "bytea[]",
5294 D::VarcharArray => "character varying[]",
5295 D::CharArray => "\"char\"[]",
5296 D::IntervalArray => "interval[]",
5297 _ => return None,
5298 })
5299}
5300
5301/// v7.37.16 — zero-allocation LIKE core: the text side walks a `&str`
5302/// cursor (char-semantic — `_` consumes one CHARACTER, `%` backtracks
5303/// only at char boundaries) instead of collecting a `Vec<char>` per
5304/// call. The old per-row collect was ~50 ns/row of pure allocator
5305/// traffic on a 50 k-row `WHERE s LIKE '%…%'` scan (the heavy.rs
5306/// like_filter 2.8× loss); the pattern side stays a compile-once
5307/// `&[char]` (see `Step::Like`).
5308pub(crate) fn like_match_str(text: &str, pat: &[char], mut pi: usize) -> Result<bool, EvalError> {
5309 let mut t = text;
5310 while pi < pat.len() {
5311 match pat[pi] {
5312 '%' => {
5313 // Collapse consecutive `%` and try every possible split.
5314 while pi < pat.len() && pat[pi] == '%' {
5315 pi += 1;
5316 }
5317 if pi == pat.len() {
5318 return Ok(true);
5319 }
5320 let mut rest = t;
5321 loop {
5322 if like_match_str(rest, pat, pi)? {
5323 return Ok(true);
5324 }
5325 match rest.chars().next() {
5326 Some(c) => rest = &rest[c.len_utf8()..],
5327 None => return Ok(false),
5328 }
5329 }
5330 }
5331 '_' => match t.chars().next() {
5332 Some(c) => {
5333 t = &t[c.len_utf8()..];
5334 pi += 1;
5335 }
5336 None => return Ok(false),
5337 },
5338 // v7.39 (round 144, like_match.c) — a trailing unpaired escape is
5339 // PG's 22025 error, but LAZILY: only when the matcher reaches it
5340 // with text left. A branch where the text is already exhausted
5341 // returns false without ever "seeing" the trailing escape
5342 // ('x' LIKE 'x\' is false; 'xy' LIKE 'x\' errors).
5343 '\\' if pi + 1 >= pat.len() => {
5344 if t.is_empty() {
5345 return Ok(false);
5346 }
5347 return Err(EvalError::TypeMismatch {
5348 detail: "LIKE pattern must not end with escape character".into(),
5349 });
5350 }
5351 '\\' => {
5352 let want = pat[pi + 1];
5353 match t.chars().next() {
5354 Some(c) if c == want => {
5355 t = &t[c.len_utf8()..];
5356 pi += 2;
5357 }
5358 _ => return Ok(false),
5359 }
5360 }
5361 c => match t.chars().next() {
5362 Some(tc) if tc == c => {
5363 t = &t[c.len_utf8()..];
5364 pi += 1;
5365 }
5366 _ => return Ok(false),
5367 },
5368 }
5369 }
5370 Ok(t.is_empty())
5371}
5372
5373/// v7.24 (round-15) — `string_to_array(text, delimiter)`.
5374fn fn_string_to_array(args: &[Value<'_>]) -> Result<Value<'static>, EvalError> {
5375 // v7.37.17 (17.6 siblings) — the 3-arg PG form adds
5376 // `null_string`: elements equal to it become SQL NULL.
5377 let (text_arg, delim_arg, null_arg) = match args {
5378 [t, d] => (t, d, None),
5379 [t, d, n] => (t, d, Some(n)),
5380 _ => {
5381 return Err(EvalError::TypeMismatch {
5382 detail: alloc::format!(
5383 "string_to_array expects 2 or 3 arguments, got {}",
5384 args.len()
5385 ),
5386 });
5387 }
5388 };
5389 let null_string: Option<&str> = match null_arg {
5390 None | Some(Value::Null) => None,
5391 Some(Value::Text(s)) => Some(s.as_ref()),
5392 Some(other) => {
5393 return Err(EvalError::TypeMismatch {
5394 detail: alloc::format!(
5395 "string_to_array null_string must be text, got {}",
5396 crate::conversions::pg_type_name_for_error_opt(other.data_type())
5397 ),
5398 });
5399 }
5400 };
5401 let text = match text_arg {
5402 Value::Null => return Ok(Value::Null),
5403 Value::Text(t) => t,
5404 other => {
5405 return Err(EvalError::TypeMismatch {
5406 detail: alloc::format!(
5407 "string_to_array expects text, got {}",
5408 crate::conversions::pg_type_name_for_error_opt(other.data_type())
5409 ),
5410 });
5411 }
5412 };
5413 // PG (9.1+): empty input → empty array, regardless of delimiter.
5414 if text.is_empty() {
5415 return Ok(Value::TextArray(Vec::new()));
5416 }
5417 let nullify = |p: String| -> Option<String> {
5418 if null_string == Some(p.as_str()) {
5419 None
5420 } else {
5421 Some(p)
5422 }
5423 };
5424 let parts: Vec<Option<String>> = match delim_arg {
5425 // NULL delimiter → one element per character.
5426 Value::Null => text.chars().map(|c| nullify(c.to_string())).collect(),
5427 Value::Text(d) if d.is_empty() => alloc::vec![nullify(text.to_string())],
5428 Value::Text(d) => text
5429 .split(d.as_ref())
5430 .map(|p| nullify(p.to_string()))
5431 .collect(),
5432 other => {
5433 return Err(EvalError::TypeMismatch {
5434 detail: alloc::format!(
5435 "string_to_array delimiter must be text, got {}",
5436 crate::conversions::pg_type_name_for_error_opt(other.data_type())
5437 ),
5438 });
5439 }
5440 };
5441 Ok(Value::TextArray(parts))
5442}
5443
5444/// v6.4.3 — `error_on_null(v)`. Returns `v` unchanged if non-NULL;
5445/// errors otherwise. Convenience to assert NOT NULL inside an
5446/// expression without wrapping it in COALESCE + raise hacks.
5447fn error_on_null(args: &[Value<'_>]) -> Result<Value<'static>, EvalError> {
5448 if args.len() != 1 {
5449 return Err(EvalError::TypeMismatch {
5450 detail: format!("error_on_null() takes 1 arg, got {}", args.len()),
5451 });
5452 }
5453 if matches!(args[0], Value::Null) {
5454 return Err(EvalError::TypeMismatch {
5455 detail: "error_on_null(): argument is NULL".into(),
5456 });
5457 }
5458 Ok(args[0].clone().into_owned())
5459}
5460
5461/// Helper: coerce a Value to an Option<String> for regex args. NULL
5462/// propagates as None (caller short-circuits to Value::Null).
5463fn text_arg(v: &Value) -> Result<Option<String>, EvalError> {
5464 match v {
5465 Value::Text(s) => Ok(Some(s.to_string())),
5466 Value::Null => Ok(None),
5467 other => Err(EvalError::TypeMismatch {
5468 detail: alloc::format!(
5469 "regex function expects TEXT arg, got {}",
5470 crate::conversions::pg_type_name_for_error_opt(other.data_type())
5471 ),
5472 }),
5473 }
5474}
5475
5476// Month-name tables shared by the date formatters in `eval::strings`
5477// (`date_format_mysql`) and `eval::datetime` via `use super::`. Kept in
5478// `eval.rs` alongside `civil_from_days` so the calendar primitives live
5479// in one place.
5480const MONTH_FULL: [&str; 12] = [
5481 "January",
5482 "February",
5483 "March",
5484 "April",
5485 "May",
5486 "June",
5487 "July",
5488 "August",
5489 "September",
5490 "October",
5491 "November",
5492 "December",
5493];
5494const MONTH_ABBR: [&str; 12] = [
5495 "Jan", "Feb", "Mar", "Apr", "May", "Jun", "Jul", "Aug", "Sep", "Oct", "Nov", "Dec",
5496];
5497
5498/// Howard Hinnant's `civil_from_days` — converts days since the Unix
5499/// epoch back to a proleptic-Gregorian (year, month, day) triple. Stays
5500/// in `eval.rs` (shared with the date SQL functions here and with
5501/// `eval::strings`); the inverse `days_from_civil` lives in
5502/// `eval::format`. Both keep the engine off `std` time facilities.
5503#[allow(clippy::cast_possible_truncation, clippy::cast_sign_loss)]
5504fn civil_from_days(days: i32) -> (i32, u32, u32) {
5505 let z = i64::from(days) + 719_468;
5506 let era = z.div_euclid(146_097);
5507 // doe ∈ [0, 146_097); fits in u32 with room to spare. Same for
5508 // every other quantity below — `as u32` truncations are safe by
5509 // construction.
5510 let doe = (z - era * 146_097) as u32;
5511 let yoe = (doe.saturating_sub(doe / 1460) + doe / 36524 - doe / 146_096) / 365;
5512 let y_base = i64::from(yoe) + era * 400;
5513 let doy = doe.saturating_sub(365 * yoe + yoe / 4 - yoe / 100);
5514 let mp = (5 * doy + 2) / 153;
5515 let d = doy.saturating_sub((153 * mp + 2) / 5) + 1;
5516 let m = if mp < 10 { mp + 3 } else { mp - 9 };
5517 let y = if m <= 2 { y_base + 1 } else { y_base };
5518 (y as i32, m, d)
5519}
5520
5521/// Add `months` (signed) to a `(year, month, day)` triple using PG's
5522/// clamp-to-last-day rule (so `'2024-01-31' + 1 month` → `'2024-02-29'`).
5523fn add_months_to_civil(y: i32, m: u32, d: u32, months: i32) -> (i32, u32, u32) {
5524 let total_months = i64::from(y) * 12 + i64::from(m) - 1 + i64::from(months);
5525 let new_year = i32::try_from(total_months.div_euclid(12)).unwrap_or(i32::MAX);
5526 let new_month_zero = total_months.rem_euclid(12);
5527 #[allow(clippy::cast_possible_truncation, clippy::cast_sign_loss)]
5528 let new_month = (new_month_zero as u32) + 1;
5529 let max_day = days_in_month(new_year, new_month);
5530 (new_year, new_month, d.min(max_day))
5531}
5532
5533const fn days_in_month(y: i32, m: u32) -> u32 {
5534 match m {
5535 1 | 3 | 5 | 7 | 8 | 10 | 12 => 31,
5536 2 => {
5537 // Proleptic Gregorian leap rule.
5538 if y.rem_euclid(4) == 0 && (y.rem_euclid(100) != 0 || y.rem_euclid(400) == 0) {
5539 29
5540 } else {
5541 28
5542 }
5543 }
5544 // 4 / 6 / 9 / 11 plus any out-of-range month (callers normalise
5545 // first, but be defensive) get the 30-day fallback.
5546 _ => 30,
5547 }
5548}
5549
5550pub(crate) fn literal_to_value(l: &Literal) -> Value<'static> {
5551 match l {
5552 Literal::Integer(n) => {
5553 if let Ok(small) = i32::try_from(*n) {
5554 Value::Int(small)
5555 } else {
5556 Value::BigInt(*n)
5557 }
5558 }
5559 Literal::Float(x) => Value::Float(*x),
5560 Literal::Numeric { unscaled, scale } => Value::Numeric {
5561 scaled: *unscaled,
5562 scale: *scale,
5563 kind: spg_storage::NumericKind::Finite,
5564 },
5565 Literal::NumericBig(s) => crate::conversions::big_literal_to_value(s),
5566 Literal::String(s) => Value::text(s.clone()),
5567 Literal::Vector(v) => Value::vector(v.clone()),
5568 Literal::TextArray(items) => Value::TextArray(items.clone()),
5569 Literal::IntArray(items) => Value::IntArray(items.clone()),
5570 Literal::BigIntArray(items) => Value::BigIntArray(items.clone()),
5571 Literal::Bool(b) => Value::Bool(*b),
5572 Literal::Null => Value::Null,
5573 Literal::Interval {
5574 months,
5575 days,
5576 micros,
5577 ..
5578 } => Value::Interval {
5579 months: *months,
5580 days: *days,
5581 micros: *micros,
5582 },
5583 }
5584}
5585
5586impl crate::Engine {
5587 /// v7.39 (read01 round 63) — run a user function whose body has its own
5588 /// FROM. The arguments are substituted into the body as literals and the
5589 /// SELECT goes through the REAL executor, so it sees exactly the rows a
5590 /// hand-written query would: the row-header visibility filter applies, and
5591 /// under in-place MVCC a dead row stays dead.
5592 ///
5593 /// PG returns the FIRST row of a scalar SQL function's body (and NULL when
5594 /// it returns none).
5595 pub(crate) fn run_user_fn_query(
5596 &self,
5597 def: &spg_storage::FunctionDef,
5598 stmt: &spg_sql::ast::SelectStatement,
5599 arg_names: &[alloc::string::String],
5600 args: &spg_storage::Row<'static>,
5601 fn_depth: u16,
5602 ) -> Result<Value<'static>, EvalError> {
5603 const MAX_QUERY_FN_DEPTH: u16 = 8;
5604 if fn_depth >= MAX_QUERY_FN_DEPTH {
5605 return Err(EvalError::TypeMismatch {
5606 detail: alloc::format!(
5607 "function {:?}: a body with its own FROM may nest at most {MAX_QUERY_FN_DEPTH} deep",
5608 def.name
5609 ),
5610 });
5611 }
5612 // Bind the arguments — the same helper the set-returning path uses, so
5613 // both resolve an argument identically (a COLUMN of the body's own FROM
5614 // shadows a same-named argument, as in PG).
5615 let owned: alloc::vec::Vec<Value<'static>> =
5616 args.values.iter().map(|v| v.clone().into_owned()).collect();
5617 let bound =
5618 bind_user_fn_args(self.active_catalog(), stmt, arg_names, &owned).map_err(|e| {
5619 EvalError::TypeMismatch {
5620 detail: alloc::format!("function {:?}: {e}", def.name),
5621 }
5622 })?;
5623
5624 // v7.39 (round 334, V55) — a SECURITY DEFINER body is authorised as
5625 // the function's OWNER. Measured on PG 18.4: a definer function
5626 // owned by `owner55` counts rows of a table `caller55` cannot read,
5627 // while the SECURITY INVOKER sibling is refused.
5628 let as_role = def.security_definer.then(|| def.owner.as_deref()).flatten();
5629 let out = self
5630 .exec_select_cancel_as(&bound, crate::CancelToken::none(), as_role)
5631 .map_err(|e| EvalError::TypeMismatch {
5632 detail: alloc::format!("function {:?}: {e}", def.name),
5633 })?;
5634 let crate::QueryResult::Rows { rows, .. } = out else {
5635 return Ok(Value::Null);
5636 };
5637 let Some(first) = rows.first() else {
5638 // No row: PG's scalar SQL function returns NULL.
5639 return Ok(Value::Null);
5640 };
5641 let v = first.values.first().cloned().unwrap_or(Value::Null);
5642 let declared = def.returns.trim();
5643 if declared.eq_ignore_ascii_case("VOID") {
5644 return Ok(Value::Null);
5645 }
5646 crate::eval::cast::cast_value(
5647 v.into_owned(),
5648 spg_sql::ast::CastTarget::Named(alloc::string::String::from(declared)),
5649 )
5650 .or_else(|_| Ok(Value::Null))
5651 }
5652}
5653
5654/// v7.39 (read01 round 65) — bind a call's arguments into a function body's
5655/// SELECT, as literals. Shared by the scalar path (round 63) and the
5656/// set-returning one, so both resolve an argument the same way — including the
5657/// rule that a COLUMN of the body's own FROM shadows a same-named argument.
5658pub(crate) fn bind_user_fn_args(
5659 cat: &spg_storage::Catalog,
5660 stmt: &spg_sql::ast::SelectStatement,
5661 arg_names: &[alloc::string::String],
5662 args: &[Value<'static>],
5663) -> Result<spg_sql::ast::SelectStatement, EvalError> {
5664 let mut bound = stmt.clone();
5665 let mut binds: alloc::collections::BTreeMap<alloc::string::String, spg_sql::ast::Expr> =
5666 alloc::collections::BTreeMap::new();
5667 for (i, name) in arg_names.iter().enumerate() {
5668 if name.is_empty() {
5669 continue;
5670 }
5671 let shadowed = body_from_tables(stmt).iter().any(|t| {
5672 cat.get(t).is_some_and(|tb| {
5673 tb.schema()
5674 .columns
5675 .iter()
5676 .any(|c| c.name.eq_ignore_ascii_case(name))
5677 })
5678 });
5679 if shadowed {
5680 continue;
5681 }
5682 let v = args.get(i).cloned().unwrap_or(Value::Null);
5683 let lit =
5684 crate::substitute::value_to_literal_expr(v).map_err(|e| EvalError::TypeMismatch {
5685 detail: alloc::format!("argument {name} cannot be bound into the body: {e}"),
5686 })?;
5687 binds.insert(name.to_ascii_lowercase(), lit);
5688 }
5689 substitute_arg_refs_in_select(&mut bound, &binds);
5690 Ok(bound)
5691}
5692
5693/// The base tables a function body's FROM names — used to decide whether an
5694/// argument name is shadowed by a column of the same name.
5695fn body_from_tables(
5696 stmt: &spg_sql::ast::SelectStatement,
5697) -> alloc::vec::Vec<alloc::string::String> {
5698 let mut out = alloc::vec::Vec::new();
5699 if let Some(from) = &stmt.from {
5700 out.push(from.primary.name.clone());
5701 for j in &from.joins {
5702 out.push(j.table.name.clone());
5703 }
5704 }
5705 out
5706}
5707
5708/// Replace every bare reference to an argument name with its literal value.
5709fn substitute_arg_refs_in_select(
5710 stmt: &mut spg_sql::ast::SelectStatement,
5711 binds: &alloc::collections::BTreeMap<alloc::string::String, spg_sql::ast::Expr>,
5712) {
5713 use spg_sql::ast::{Expr, SelectItem};
5714 fn walk(e: &mut Expr, binds: &alloc::collections::BTreeMap<alloc::string::String, Expr>) {
5715 match e {
5716 Expr::Column(c) => {
5717 if c.qualifier.is_none()
5718 && let Some(lit) = binds.get(&c.name.to_ascii_lowercase())
5719 {
5720 *e = lit.clone();
5721 }
5722 }
5723 Expr::Binary { lhs, rhs, .. } => {
5724 walk(lhs, binds);
5725 walk(rhs, binds);
5726 }
5727 Expr::Unary { expr, .. } | Expr::Cast { expr, .. } => walk(expr, binds),
5728 Expr::FunctionCall { args, .. } => args.iter_mut().for_each(|a| walk(a, binds)),
5729 Expr::Case {
5730 operand,
5731 branches,
5732 else_branch,
5733 } => {
5734 if let Some(o) = operand {
5735 walk(o, binds);
5736 }
5737 for (c, v) in branches.iter_mut() {
5738 walk(c, binds);
5739 walk(v, binds);
5740 }
5741 if let Some(x) = else_branch {
5742 walk(x, binds);
5743 }
5744 }
5745 Expr::InList { expr, list, .. } => {
5746 walk(expr, binds);
5747 list.iter_mut().for_each(|it| walk(it, binds));
5748 }
5749 Expr::AnyAll { expr, array, .. } => {
5750 walk(expr, binds);
5751 walk(array, binds);
5752 }
5753 Expr::Array(items) => items.iter_mut().for_each(|it| walk(it, binds)),
5754 Expr::ArraySubscript { target, index } => {
5755 walk(target, binds);
5756 walk(index, binds);
5757 }
5758 _ => {}
5759 }
5760 }
5761 for item in &mut stmt.items {
5762 if let SelectItem::Expr { expr, .. } = item {
5763 walk(expr, binds);
5764 }
5765 }
5766 if let Some(w) = &mut stmt.where_ {
5767 walk(w, binds);
5768 }
5769 if let Some(h) = &mut stmt.having {
5770 walk(h, binds);
5771 }
5772 if let Some(gs) = &mut stmt.group_by {
5773 gs.iter_mut().for_each(|g| walk(g, binds));
5774 }
5775 for o in &mut stmt.order_by {
5776 walk(&mut o.expr, binds);
5777 }
5778 if let Some(from) = &mut stmt.from {
5779 for j in &mut from.joins {
5780 if let Some(on) = &mut j.on {
5781 walk(on, binds);
5782 }
5783 }
5784 }
5785 // v7.39 (read01 round 69) — the UNION peers. A body like
5786 // `SELECT k UNION ALL SELECT k * 10` has its second half in `unions`, and
5787 // leaving it unsubstituted made the argument look like a missing column.
5788 for (_, peer) in &mut stmt.unions {
5789 substitute_arg_refs_in_select(peer, binds);
5790 }
5791 for cte in &mut stmt.ctes {
5792 if let Some(s) = cte.body.as_select_mut() {
5793 substitute_arg_refs_in_select(s, binds);
5794 }
5795 }
5796}
5797
5798impl crate::Engine {
5799 /// v7.39 (read01 round 64) — call a plpgsql function as a scalar. The body
5800 /// runs on the interpreter the DO block and triggers already use, with the
5801 /// arguments bound as locals; its `SELECT … INTO` and `FOR … IN SELECT`
5802 /// resolvers go through the READ path, so what the body sees is what a
5803 /// hand-written query would see (visibility filter and all).
5804 ///
5805 /// A body that writes is refused — the call arrives through expression
5806 /// evaluation, which holds the engine immutably. Refusing is the honest
5807 /// answer; silently dropping the write would be the worst one.
5808 pub(crate) fn call_plpgsql_scalar_fn(
5809 &self,
5810 def: &spg_storage::FunctionDef,
5811 arg_names: &[alloc::string::String],
5812 args: &spg_storage::Row<'static>,
5813 ) -> Result<Value<'static>, EvalError> {
5814 let block =
5815 spg_sql::parse_function_body(def.body.trim()).map_err(|e| EvalError::TypeMismatch {
5816 detail: alloc::format!("function {:?} body does not parse: {e}", def.name),
5817 })?;
5818 let mut locals: alloc::collections::BTreeMap<alloc::string::String, Value<'static>> =
5819 alloc::collections::BTreeMap::new();
5820 for (i, n) in arg_names.iter().enumerate() {
5821 if n.is_empty() {
5822 continue;
5823 }
5824 locals.insert(
5825 n.to_ascii_lowercase(),
5826 args.values.get(i).cloned().unwrap_or(Value::Null),
5827 );
5828 }
5829 let dts = self
5830 .session_param("default_text_search_config")
5831 .map(alloc::string::String::from);
5832
5833 let select_into = |stmt: &spg_sql::ast::Statement| -> Result<
5834 Value<'static>,
5835 crate::triggers::TriggerError,
5836 > {
5837 let spg_sql::ast::Statement::Select(s) = stmt else {
5838 return Err(crate::triggers::TriggerError::EvalFailed {
5839 function: def.name.clone(),
5840 cause: EvalError::TypeMismatch {
5841 detail: "SELECT … INTO body must be a SELECT".into(),
5842 },
5843 });
5844 };
5845 let r = self
5846 .exec_select_cancel(s, crate::CancelToken::none())
5847 .map_err(|e| crate::triggers::TriggerError::EvalFailed {
5848 function: def.name.clone(),
5849 cause: EvalError::TypeMismatch {
5850 detail: alloc::format!("SELECT … INTO failed: {e}"),
5851 },
5852 })?;
5853 match r {
5854 crate::QueryResult::Rows { rows, .. } => Ok(rows
5855 .into_iter()
5856 .next()
5857 .and_then(|row| row.values.into_iter().next())
5858 .unwrap_or(Value::Null)),
5859 _ => Ok(Value::Null),
5860 }
5861 };
5862 let for_query = |stmt: &spg_sql::ast::Statement| -> Result<
5863 (
5864 alloc::vec::Vec<alloc::string::String>,
5865 alloc::vec::Vec<alloc::vec::Vec<Value<'static>>>,
5866 ),
5867 crate::triggers::TriggerError,
5868 > {
5869 let spg_sql::ast::Statement::Select(s) = stmt else {
5870 return Err(crate::triggers::TriggerError::EvalFailed {
5871 function: def.name.clone(),
5872 cause: EvalError::TypeMismatch {
5873 detail: "FOR … IN body must be a SELECT".into(),
5874 },
5875 });
5876 };
5877 let r = self
5878 .exec_select_cancel(s, crate::CancelToken::none())
5879 .map_err(|e| crate::triggers::TriggerError::EvalFailed {
5880 function: def.name.clone(),
5881 cause: EvalError::TypeMismatch {
5882 detail: alloc::format!("FOR … IN SELECT failed: {e}"),
5883 },
5884 })?;
5885 match r {
5886 crate::QueryResult::Rows { columns, rows } => Ok((
5887 columns.iter().map(|c| c.name.clone()).collect(),
5888 rows.into_iter().map(|row| row.values).collect(),
5889 )),
5890 _ => Ok((alloc::vec::Vec::new(), alloc::vec::Vec::new())),
5891 }
5892 };
5893
5894 let out = crate::triggers::call_plpgsql_scalar(
5895 &def.name,
5896 &block,
5897 locals,
5898 dts.as_deref(),
5899 Some(&select_into),
5900 Some(&for_query),
5901 // A scalar call has no set to build; RETURN NEXT / RETURN QUERY are
5902 // errors here, as in PG.
5903 // (second None below: the read path holds the engine immutably, so
5904 // RAISE messages have nowhere session-bound to go — B3 residual.)
5905 None,
5906 None,
5907 )
5908 .map_err(|e| EvalError::TypeMismatch {
5909 detail: alloc::format!("{e}"),
5910 })?;
5911 let declared = def.returns.trim();
5912 let Some(v) = out else {
5913 if declared.eq_ignore_ascii_case("VOID") {
5914 return Ok(Value::Null);
5915 }
5916 // PG: a non-void function that falls out of the bottom.
5917 return Err(EvalError::TypeMismatch {
5918 detail: alloc::format!(
5919 "control reached end of function {:?} without RETURN",
5920 def.name
5921 ),
5922 });
5923 };
5924 if declared.eq_ignore_ascii_case("VOID") {
5925 return Ok(Value::Null);
5926 }
5927 crate::eval::cast::cast_value(
5928 v.into_owned(),
5929 spg_sql::ast::CastTarget::Named(alloc::string::String::from(declared)),
5930 )
5931 .or_else(|_| Ok(Value::Null))
5932 }
5933}
5934
5935impl crate::Engine {
5936 /// v7.39 (read01 round 66) — run a `RETURNS SETOF` plpgsql function and
5937 /// collect the rows `RETURN NEXT` / `RETURN QUERY` appended. Same
5938 /// interpreter, same read-path resolvers as the scalar call — the only
5939 /// difference is that a SINK is provided, which is what makes those two
5940 /// statements legal.
5941 pub(crate) fn call_plpgsql_setof_fn(
5942 &self,
5943 def: &spg_storage::FunctionDef,
5944 arg_names: &[alloc::string::String],
5945 args: &[Value<'static>],
5946 ) -> Result<alloc::vec::Vec<alloc::vec::Vec<Value<'static>>>, EvalError> {
5947 let block =
5948 spg_sql::parse_function_body(def.body.trim()).map_err(|e| EvalError::TypeMismatch {
5949 detail: alloc::format!("function {:?} body does not parse: {e}", def.name),
5950 })?;
5951 let mut locals: alloc::collections::BTreeMap<alloc::string::String, Value<'static>> =
5952 alloc::collections::BTreeMap::new();
5953 for (i, n) in arg_names.iter().enumerate() {
5954 if n.is_empty() {
5955 continue;
5956 }
5957 locals.insert(
5958 n.to_ascii_lowercase(),
5959 args.get(i).cloned().unwrap_or(Value::Null),
5960 );
5961 }
5962 let dts = self
5963 .session_param("default_text_search_config")
5964 .map(alloc::string::String::from);
5965 let run_select = |stmt: &spg_sql::ast::Statement,
5966 what: &str|
5967 -> Result<crate::QueryResult, crate::triggers::TriggerError> {
5968 let spg_sql::ast::Statement::Select(s) = stmt else {
5969 return Err(crate::triggers::TriggerError::EvalFailed {
5970 function: def.name.clone(),
5971 cause: EvalError::TypeMismatch {
5972 detail: alloc::format!("{what} body must be a SELECT"),
5973 },
5974 });
5975 };
5976 self.exec_select_cancel(s, crate::CancelToken::none())
5977 .map_err(|e| crate::triggers::TriggerError::EvalFailed {
5978 function: def.name.clone(),
5979 cause: EvalError::TypeMismatch {
5980 detail: alloc::format!("{what} failed: {e}"),
5981 },
5982 })
5983 };
5984 let select_into = |stmt: &spg_sql::ast::Statement| -> Result<
5985 Value<'static>,
5986 crate::triggers::TriggerError,
5987 > {
5988 match run_select(stmt, "SELECT … INTO")? {
5989 crate::QueryResult::Rows { rows, .. } => Ok(rows
5990 .into_iter()
5991 .next()
5992 .and_then(|r| r.values.into_iter().next())
5993 .unwrap_or(Value::Null)),
5994 _ => Ok(Value::Null),
5995 }
5996 };
5997 let for_query = |stmt: &spg_sql::ast::Statement| -> Result<
5998 (
5999 alloc::vec::Vec<alloc::string::String>,
6000 alloc::vec::Vec<alloc::vec::Vec<Value<'static>>>,
6001 ),
6002 crate::triggers::TriggerError,
6003 > {
6004 match run_select(stmt, "FOR … IN / RETURN QUERY")? {
6005 crate::QueryResult::Rows { columns, rows } => Ok((
6006 columns.iter().map(|c| c.name.clone()).collect(),
6007 rows.into_iter().map(|r| r.values).collect(),
6008 )),
6009 _ => Ok((alloc::vec::Vec::new(), alloc::vec::Vec::new())),
6010 }
6011 };
6012 let sink: core::cell::RefCell<alloc::vec::Vec<alloc::vec::Vec<Value<'static>>>> =
6013 core::cell::RefCell::new(alloc::vec::Vec::new());
6014 crate::triggers::call_plpgsql_scalar(
6015 &def.name,
6016 &block,
6017 locals,
6018 dts.as_deref(),
6019 Some(&select_into),
6020 Some(&for_query),
6021 Some(&sink),
6022 // Read path — immutable engine borrow; B3 residual.
6023 None,
6024 )
6025 .map_err(|e| EvalError::TypeMismatch {
6026 detail: alloc::format!("{e}"),
6027 })?;
6028 Ok(sink.into_inner())
6029 }
6030}
6031
6032/// v7.39 (round 236) — resolve an ARRAY constructor's element types the way
6033/// PG does. Typed elements must share a type category; a bare string or NULL
6034/// literal is untyped and converts to whatever the typed elements resolved
6035/// to, reporting the value (not a type mismatch) when it will not convert.
6036fn unify_array_elements(
6037 items: &[Expr],
6038 materialised: &mut [Value<'static>],
6039) -> Result<(), EvalError> {
6040 unify_construct_values("ARRAY", items, materialised)
6041}
6042
6043/// v7.39 (round 238) — an `IN (...)` list must be comparable with its
6044/// needle. Reports the operator the way PG does
6045/// ("operator does not exist: integer = text"), and — like round 237 —
6046/// judges only the operands whose type is genuinely known.
6047fn require_in_list_comparable(
6048 needle: &Expr,
6049 list: &[Expr],
6050 ctx: &EvalContext<'_>,
6051) -> Result<(), EvalError> {
6052 let known_ty = |e: &Expr| {
6053 matches!(e, Expr::Cast { .. } | Expr::Literal(_) | Expr::Column(_))
6054 .then(|| crate::describe::describe_expr(e, ctx.columns).map(|s| s.ty))
6055 .flatten()
6056 };
6057 // An untyped literal adopts the needle's type, so it never conflicts.
6058 //
6059 // v7.39 (round 652) — and so does a cast to one of the reg types.
6060 // `'pg_class'::regclass` carries an oid AND a name, which is why
6061 // `compare` has arms for both, but it DESCRIBES as text — SPG has no
6062 // `DataType` for it. This check ran before any comparison and
6063 // refused `WHERE oid IN ('pg_class'::regclass, …)` as `bigint =
6064 // text`, while the identical `oid = 'pg_class'::regclass` worked.
6065 // That shape is how pg_dump, ORMs and monitoring queries name a
6066 // handful of relations at once, so it is not a corner.
6067 let reg_cast = |e: &Expr| {
6068 match e {
6069 Expr::Cast { target, .. } => match target {
6070 spg_sql::ast::CastTarget::RegClass | spg_sql::ast::CastTarget::RegType => true,
6071 // `regproc` and `regnamespace` have no variant of their
6072 // own; they arrive through the generic named path.
6073 spg_sql::ast::CastTarget::Named(n) => {
6074 n.eq_ignore_ascii_case("regproc")
6075 || n.eq_ignore_ascii_case("regnamespace")
6076 || n.eq_ignore_ascii_case("regtype")
6077 || n.eq_ignore_ascii_case("regclass")
6078 }
6079 _ => false,
6080 },
6081 _ => false,
6082 }
6083 };
6084 let untyped = |e: &Expr| {
6085 reg_cast(e)
6086 || matches!(
6087 e,
6088 Expr::Literal(spg_sql::ast::Literal::String(_))
6089 | Expr::Literal(spg_sql::ast::Literal::Null)
6090 )
6091 };
6092 // The needle can be untyped too (`NULL IN (1,2)`): SPG has no `Unknown`
6093 // DataType, so a bare NULL describes as TEXT and would look like a text
6094 // needle conflicting with integer list items.
6095 if untyped(needle) {
6096 return Ok(());
6097 }
6098 let Some(nt) = known_ty(needle) else {
6099 return Ok(());
6100 };
6101 for item in list {
6102 if untyped(item) {
6103 continue;
6104 }
6105 let Some(it) = known_ty(item) else { continue };
6106 if !crate::conversions::types_unify(nt, it) {
6107 return Err(EvalError::TypeMismatch {
6108 detail: alloc::format!(
6109 "operator does not exist: {} = {}",
6110 crate::conversions::pg_type_name_for_error(nt),
6111 crate::conversions::pg_type_name_for_error(it),
6112 ),
6113 });
6114 }
6115 }
6116 Ok(())
6117}
6118
6119/// v7.39 (round 237) — STATIC branch-type resolution for the constructs
6120/// whose branches must not all be evaluated: CASE runs only the branch it
6121/// takes, and a COALESCE / GREATEST argument may have side effects
6122/// (`COALESCE(nextval('s'), 1)`), so the check reads each branch's declared
6123/// type instead of its value. Same rule and wording as the value-driven
6124/// ARRAY path below; an untyped literal is converted here (a literal has no
6125/// side effects) so a value that will not convert is reported as PG does.
6126/// v7.39 (round 609) — takes anything that yields the branches, so the
6127/// COALESCE caller no longer builds a `Vec<&Expr>` of them for every row.
6128pub(crate) fn unify_branch_types_static<'e>(
6129 construct: &str,
6130 branches: impl IntoIterator<Item = &'e Expr> + Clone,
6131 ctx: &EvalContext<'_>,
6132) -> Result<(), EvalError> {
6133 use spg_storage::DataType;
6134 let untyped = |e: &Expr| {
6135 matches!(
6136 e,
6137 Expr::Literal(spg_sql::ast::Literal::String(_))
6138 | Expr::Literal(spg_sql::ast::Literal::Null)
6139 )
6140 };
6141 let mut resolved: Option<DataType> = None;
6142 for e in branches.clone() {
6143 if untyped(e) {
6144 continue;
6145 }
6146 // Only branches whose type is GENUINELY known take part. A general
6147 // `describe_expr` is a best-effort hint for wire type tags, not a
6148 // type checker: it reports a binary operator as its left operand's
6149 // type, so `payload->'a'` (jsonb in PG) came back as text and this
6150 // check refused a working `COALESCE(payload->'a', '{}'::jsonb)`.
6151 // Refusing a valid query is worse than missing an invalid one, so
6152 // the check confines itself to an explicit cast, a typed literal and
6153 // a plain column reference.
6154 let known = matches!(e, Expr::Cast { .. } | Expr::Literal(_) | Expr::Column(_));
6155 if !known {
6156 continue;
6157 }
6158 let Some(ty) = crate::describe::describe_expr_type(e, ctx.columns) else {
6159 continue;
6160 };
6161 match resolved {
6162 None => resolved = Some(ty),
6163 Some(prev) if crate::conversions::types_unify(prev, ty) => {
6164 if matches!(prev, DataType::Int | DataType::SmallInt) {
6165 resolved = Some(ty);
6166 }
6167 }
6168 Some(prev) => {
6169 return Err(EvalError::TypeMismatch {
6170 detail: alloc::format!(
6171 "{construct} types {} and {} cannot be matched",
6172 crate::conversions::pg_type_name_for_error(prev),
6173 crate::conversions::pg_type_name_for_error(ty),
6174 ),
6175 });
6176 }
6177 }
6178 }
6179 let Some(target) = resolved else {
6180 return Ok(());
6181 };
6182 if matches!(target, DataType::Text) {
6183 return Ok(());
6184 }
6185 // v7.39 (round 398) — MySQL aggregates a mixed int/string CASE /
6186 // COALESCE to a string (`CASE WHEN 1 THEN 1 ELSE 'x' END` is '1', not an
6187 // error); PG requires the untyped string literals to coerce to the
6188 // resolved numeric type, so it refuses. Under the dialect, skip that
6189 // coercion check — the value is returned as-is / widened by the caller.
6190 if ctx.mysql_dialect {
6191 return Ok(());
6192 }
6193 for e in branches {
6194 if !untyped(e) {
6195 continue;
6196 }
6197 if let Expr::Literal(spg_sql::ast::Literal::String(lit)) = e {
6198 crate::conversions::coerce_value(Value::text(lit.clone()), target, "", 0).map_err(
6199 |err| match err {
6200 crate::EngineError::Eval(ev) => ev,
6201 other => EvalError::TypeMismatch {
6202 detail: alloc::format!("{other}"),
6203 },
6204 },
6205 )?;
6206 }
6207 }
6208 Ok(())
6209}
6210
6211/// v7.39 (round 237) — the same resolution for every construct that builds
6212/// one value out of several branches: ARRAY, CASE, COALESCE, GREATEST and
6213/// LEAST. PG names the construct in the message ("CASE types text and
6214/// integer cannot be matched"), which is why the caller passes it in.
6215pub(crate) fn unify_construct_values(
6216 construct: &str,
6217 items: &[Expr],
6218 materialised: &mut [Value<'static>],
6219) -> Result<(), EvalError> {
6220 use spg_storage::DataType;
6221 let untyped = |e: &Expr| {
6222 matches!(
6223 e,
6224 Expr::Literal(spg_sql::ast::Literal::String(_))
6225 | Expr::Literal(spg_sql::ast::Literal::Null)
6226 )
6227 };
6228 // The type the typed elements agree on, if any.
6229 let mut resolved: Option<DataType> = None;
6230 for (i, v) in materialised.iter().enumerate() {
6231 if items.get(i).is_some_and(untyped) {
6232 continue;
6233 }
6234 let Some(ty) = v.data_type() else { continue };
6235 match resolved {
6236 None => resolved = Some(ty),
6237 Some(prev) if crate::conversions::types_unify(prev, ty) => {
6238 // Keep the wider of the two so the coercion below targets it.
6239 if matches!(prev, DataType::Int | DataType::SmallInt) {
6240 resolved = Some(ty);
6241 }
6242 }
6243 Some(prev) => {
6244 return Err(EvalError::TypeMismatch {
6245 detail: alloc::format!(
6246 "{construct} types {} and {} cannot be matched",
6247 crate::conversions::pg_type_name_for_error(prev),
6248 crate::conversions::pg_type_name_for_error(ty),
6249 ),
6250 });
6251 }
6252 }
6253 }
6254 // Untyped literals adopt that type; a failure names the value, as PG does.
6255 let Some(target) = resolved else {
6256 return Ok(());
6257 };
6258 if matches!(target, DataType::Text) {
6259 return Ok(());
6260 }
6261 for (i, v) in materialised.iter_mut().enumerate() {
6262 if !items.get(i).is_some_and(untyped) || matches!(v, Value::Null) {
6263 continue;
6264 }
6265 *v = crate::conversions::coerce_value(v.clone(), target, "", i).map_err(|e| match e {
6266 crate::EngineError::Eval(ev) => ev,
6267 other => EvalError::TypeMismatch {
6268 detail: alloc::format!("{other}"),
6269 },
6270 })?;
6271 }
6272 Ok(())
6273}
6274
6275/// v7.39 (round 309, V30) — split `'<timestamp> <zone name>'` into the
6276/// wall-clock reading and the zone token, for the zone-less target types.
6277///
6278/// Returns `None` when the literal parses on its own (nothing to strip)
6279/// or when the trailing token is not zone-SHAPED — those keep the
6280/// ordinary "invalid input syntax" path, which is what PG answers for
6281/// `'2020-01-01 10:00:00 xyz'`. Whether a zone-shaped token is a REAL
6282/// zone is the caller's question; getting that wrong is a different
6283/// error in PG, and conflating the two would report a malformed literal
6284/// for a merely-misspelled zone.
6285///
6286/// Deliberately does not accept a bare time (`'10:00:00 America/New_York'`):
6287/// PG refuses a named zone there, and only reaches this spelling through
6288/// a full timestamp literal.
6289fn split_trailing_zone_name(txt: &str, order: format::DateOrder) -> Option<(i64, &str)> {
6290 // Already valid without help — leave it alone.
6291 if format::parse_timestamp_literal_wall_ordered(txt, order).is_some() {
6292 return None;
6293 }
6294 let trimmed = txt.trim_end();
6295 let idx = trimmed.rfind(' ')?;
6296 let (head, tail) = (trimmed[..idx].trim(), trimmed[idx + 1..].trim());
6297 // An era marker is part of the timestamp, not a zone.
6298 let zone_shaped = tail.len() > 1
6299 && tail.bytes().any(|b| b.is_ascii_alphabetic())
6300 && !tail.eq_ignore_ascii_case("bc")
6301 && !tail.eq_ignore_ascii_case("ad");
6302 if !zone_shaped {
6303 return None;
6304 }
6305 let wall = format::parse_timestamp_literal_wall_ordered(head, order)?;
6306 Some((wall, tail))
6307}
6308
6309#[cfg(test)]
6310mod tests {
6311 use super::*;
6312 use alloc::vec;
6313 use spg_sql::ast::UnOp;
6314 use spg_storage::{ColumnSchema, DataType, Row};
6315
6316 fn col(name: &str, ty: DataType) -> ColumnSchema {
6317 ColumnSchema::new(name, ty, true)
6318 }
6319
6320 fn ctx<'a>(cols: &'a [ColumnSchema], alias: Option<&'a str>) -> EvalContext<'a> {
6321 EvalContext::new(cols, alias)
6322 }
6323
6324 /// v7.32 (P4 borrow channel) differential: the borrowed comparison
6325 /// fast path in `eval_expr`'s Binary arm must be byte-for-byte the
6326 /// pre-P4 owned path (`apply_binary` on cloned operands) across a
6327 /// cross-type value matrix and every comparison operator — covering
6328 /// the fast-path types (Text/Int/Float/Date/Timestamp/Bool/Null) and
6329 /// the owned-fallback types (Numeric/Interval).
6330 #[test]
6331 fn borrowed_compare_equals_owned_apply_binary() {
6332 let vals = vec![
6333 Value::Null,
6334 Value::Bool(true),
6335 Value::Bool(false),
6336 Value::SmallInt(3),
6337 Value::Int(3),
6338 Value::Int(-1),
6339 Value::BigInt(3),
6340 Value::BigInt(100),
6341 Value::Float(3.0),
6342 Value::Float(2.5),
6343 Value::text(String::new()),
6344 Value::text("a"),
6345 Value::text("b"),
6346 Value::Date(10),
6347 Value::Timestamp(1000),
6348 Value::Numeric {
6349 scaled: 30,
6350 scale: 1,
6351 kind: spg_storage::NumericKind::Finite,
6352 },
6353 Value::Interval {
6354 months: 0,
6355 days: 0,
6356 micros: 5,
6357 },
6358 ];
6359 let ops = [
6360 BinOp::Eq,
6361 BinOp::NotEq,
6362 BinOp::Lt,
6363 BinOp::LtEq,
6364 BinOp::Gt,
6365 BinOp::GtEq,
6366 ];
6367 let cs = vec![col("x", DataType::Int), col("y", DataType::Int)];
6368 let c = ctx(&cs, None);
6369 let lhs = Expr::Column(ColumnName {
6370 qualifier: None,
6371 name: "x".into(),
6372 });
6373 let rhs = Expr::Column(ColumnName {
6374 qualifier: None,
6375 name: "y".into(),
6376 });
6377 for l in &vals {
6378 for r in &vals {
6379 let row = Row::new(vec![l.clone(), r.clone()]);
6380 for op in ops {
6381 let got = eval_expr(
6382 &Expr::Binary {
6383 lhs: alloc::boxed::Box::new(lhs.clone()),
6384 op,
6385 rhs: alloc::boxed::Box::new(rhs.clone()),
6386 },
6387 &row,
6388 &c,
6389 );
6390 // Pre-P4 reference: owned operands through apply_binary
6391 // (collation fold is a no-op for non-CI columns).
6392 let want = apply_binary(op, l.clone(), r.clone());
6393 assert_eq!(
6394 format!("{got:?}"),
6395 format!("{want:?}"),
6396 "op={op:?} l={l:?} r={r:?}"
6397 );
6398 }
6399 }
6400 }
6401 }
6402
6403 fn lit(n: i64) -> Expr {
6404 Expr::Literal(Literal::Integer(n))
6405 }
6406
6407 fn null() -> Expr {
6408 Expr::Literal(Literal::Null)
6409 }
6410
6411 fn col_ref(name: &str) -> Expr {
6412 Expr::Column(ColumnName {
6413 qualifier: None,
6414 name: name.into(),
6415 })
6416 }
6417
6418 #[test]
6419 fn literal_evaluates_to_value() {
6420 let r = Row::new(vec![]);
6421 let cs: [ColumnSchema; 0] = [];
6422 let c = ctx(&cs, None);
6423 assert_eq!(eval_expr(&lit(42), &r, &c).unwrap(), Value::Int(42));
6424 assert_eq!(
6425 eval_expr(&Expr::Literal(Literal::Float(1.5)), &r, &c).unwrap(),
6426 Value::Float(1.5)
6427 );
6428 assert_eq!(eval_expr(&null(), &r, &c).unwrap(), Value::Null);
6429 }
6430
6431 #[test]
6432 fn column_lookup_unqualified() {
6433 let cs = vec![col("a", DataType::Int), col("b", DataType::Text)];
6434 let r = Row::new(vec![Value::Int(7), Value::text("hi")]);
6435 let c = ctx(&cs, None);
6436 assert_eq!(eval_expr(&col_ref("a"), &r, &c).unwrap(), Value::Int(7));
6437 assert_eq!(eval_expr(&col_ref("b"), &r, &c).unwrap(), Value::text("hi"));
6438 }
6439
6440 #[test]
6441 fn column_not_found_errors() {
6442 let cs = vec![col("a", DataType::Int)];
6443 let r = Row::new(vec![Value::Int(0)]);
6444 let c = ctx(&cs, None);
6445 let err = eval_expr(&col_ref("ghost"), &r, &c).unwrap_err();
6446 assert!(matches!(err, EvalError::ColumnNotFound { ref name } if name == "ghost"));
6447 }
6448
6449 #[test]
6450 fn qualified_column_matches_alias() {
6451 let cs = vec![col("a", DataType::Int)];
6452 let r = Row::new(vec![Value::Int(5)]);
6453 let c = ctx(&cs, Some("u"));
6454 let qualified = Expr::Column(ColumnName {
6455 qualifier: Some("u".into()),
6456 name: "a".into(),
6457 });
6458 assert_eq!(eval_expr(&qualified, &r, &c).unwrap(), Value::Int(5));
6459 }
6460
6461 #[test]
6462 fn qualified_column_unknown_alias_errors() {
6463 let cs = vec![col("a", DataType::Int)];
6464 let r = Row::new(vec![Value::Int(5)]);
6465 let c = ctx(&cs, Some("u"));
6466 let wrong = Expr::Column(ColumnName {
6467 qualifier: Some("x".into()),
6468 name: "a".into(),
6469 });
6470 assert!(matches!(
6471 eval_expr(&wrong, &r, &c).unwrap_err(),
6472 EvalError::UnknownQualifier { .. }
6473 ));
6474 }
6475
6476 #[test]
6477 fn arithmetic_with_widening() {
6478 let r = Row::new(vec![]);
6479 let cs: [ColumnSchema; 0] = [];
6480 let c = ctx(&cs, None);
6481 let e = Expr::Binary {
6482 lhs: alloc::boxed::Box::new(lit(2)),
6483 op: BinOp::Add,
6484 rhs: alloc::boxed::Box::new(Expr::Literal(Literal::Float(0.5))),
6485 };
6486 assert_eq!(eval_expr(&e, &r, &c).unwrap(), Value::Float(2.5));
6487 }
6488
6489 #[test]
6490 fn division_by_zero_errors() {
6491 let r = Row::new(vec![]);
6492 let cs: [ColumnSchema; 0] = [];
6493 let c = ctx(&cs, None);
6494 let e = Expr::Binary {
6495 lhs: alloc::boxed::Box::new(lit(1)),
6496 op: BinOp::Div,
6497 rhs: alloc::boxed::Box::new(lit(0)),
6498 };
6499 assert_eq!(
6500 eval_expr(&e, &r, &c).unwrap_err(),
6501 EvalError::DivisionByZero
6502 );
6503 }
6504
6505 #[test]
6506 fn comparison_returns_bool() {
6507 let r = Row::new(vec![]);
6508 let cs: [ColumnSchema; 0] = [];
6509 let c = ctx(&cs, None);
6510 let e = Expr::Binary {
6511 lhs: alloc::boxed::Box::new(lit(1)),
6512 op: BinOp::Lt,
6513 rhs: alloc::boxed::Box::new(lit(2)),
6514 };
6515 assert_eq!(eval_expr(&e, &r, &c).unwrap(), Value::Bool(true));
6516 }
6517
6518 #[test]
6519 fn null_propagates_through_arithmetic() {
6520 let r = Row::new(vec![]);
6521 let cs: [ColumnSchema; 0] = [];
6522 let c = ctx(&cs, None);
6523 let e = Expr::Binary {
6524 lhs: alloc::boxed::Box::new(lit(1)),
6525 op: BinOp::Add,
6526 rhs: alloc::boxed::Box::new(null()),
6527 };
6528 assert_eq!(eval_expr(&e, &r, &c).unwrap(), Value::Null);
6529 }
6530
6531 #[test]
6532 fn stack_depth_guard_trips_on_pathological_nesting() {
6533 // Built directly as an AST — the parser's own budgets (256
6534 // chained binary operators, 64 nesting levels) reject such SQL
6535 // long before eval sees it, so this exercises the eval-side
6536 // guard on its own. 30 000 frames overshoot the 768 KiB budget
6537 // at any conceivable frame size; the guard errors at the byte
6538 // budget, far below the worker stack, so deeper is safer.
6539 let mut e = Expr::Literal(Literal::Bool(true));
6540 for _ in 0..30_000 {
6541 e = Expr::Binary {
6542 lhs: alloc::boxed::Box::new(e),
6543 op: BinOp::And,
6544 rhs: alloc::boxed::Box::new(Expr::Literal(Literal::Bool(true))),
6545 };
6546 }
6547 let r = Row::new(vec![]);
6548 let cs: [ColumnSchema; 0] = [];
6549 let c = ctx(&cs, None);
6550 let err = eval_expr(&e, &r, &c).unwrap_err();
6551 assert!(matches!(err, EvalError::StackDepthExceeded), "{err:?}");
6552 // Dropping a 30 000-deep Box chain recurses in the drop glue —
6553 // deeper than the eval guard allows the EVAL side to go — so
6554 // leak it rather than gamble on the test thread's stack.
6555 core::mem::forget(e);
6556 }
6557
6558 #[test]
6559 fn and_three_valued_logic() {
6560 let r = Row::new(vec![]);
6561 let cs: [ColumnSchema; 0] = [];
6562 let c = ctx(&cs, None);
6563 let tt = |a: bool, b_null: bool| Expr::Binary {
6564 lhs: alloc::boxed::Box::new(Expr::Literal(Literal::Bool(a))),
6565 op: BinOp::And,
6566 rhs: alloc::boxed::Box::new(if b_null {
6567 null()
6568 } else {
6569 Expr::Literal(Literal::Bool(true))
6570 }),
6571 };
6572 // FALSE AND NULL → FALSE
6573 assert_eq!(
6574 eval_expr(&tt(false, true), &r, &c).unwrap(),
6575 Value::Bool(false)
6576 );
6577 // TRUE AND NULL → NULL
6578 assert_eq!(eval_expr(&tt(true, true), &r, &c).unwrap(), Value::Null);
6579 // TRUE AND TRUE → TRUE
6580 assert_eq!(
6581 eval_expr(&tt(true, false), &r, &c).unwrap(),
6582 Value::Bool(true)
6583 );
6584 }
6585
6586 #[test]
6587 fn or_three_valued_logic() {
6588 let r = Row::new(vec![]);
6589 let cs: [ColumnSchema; 0] = [];
6590 let c = ctx(&cs, None);
6591 let or_with_null = |a: bool| Expr::Binary {
6592 lhs: alloc::boxed::Box::new(Expr::Literal(Literal::Bool(a))),
6593 op: BinOp::Or,
6594 rhs: alloc::boxed::Box::new(null()),
6595 };
6596 // TRUE OR NULL → TRUE
6597 assert_eq!(
6598 eval_expr(&or_with_null(true), &r, &c).unwrap(),
6599 Value::Bool(true)
6600 );
6601 // FALSE OR NULL → NULL
6602 assert_eq!(
6603 eval_expr(&or_with_null(false), &r, &c).unwrap(),
6604 Value::Null
6605 );
6606 }
6607
6608 #[test]
6609 fn not_on_null_is_null() {
6610 let r = Row::new(vec![]);
6611 let cs: [ColumnSchema; 0] = [];
6612 let c = ctx(&cs, None);
6613 let e = Expr::Unary {
6614 op: UnOp::Not,
6615 expr: alloc::boxed::Box::new(null()),
6616 };
6617 assert_eq!(eval_expr(&e, &r, &c).unwrap(), Value::Null);
6618 }
6619
6620 #[test]
6621 fn text_comparison_lexicographic() {
6622 let r = Row::new(vec![]);
6623 let cs: [ColumnSchema; 0] = [];
6624 let c = ctx(&cs, None);
6625 let e = Expr::Binary {
6626 lhs: alloc::boxed::Box::new(Expr::Literal(Literal::String("apple".into()))),
6627 op: BinOp::Lt,
6628 rhs: alloc::boxed::Box::new(Expr::Literal(Literal::String("banana".into()))),
6629 };
6630 assert_eq!(eval_expr(&e, &r, &c).unwrap(), Value::Bool(true));
6631 }
6632
6633 #[test]
6634 fn interval_format_basics() {
6635 // v7.37.5 β — three-arg signature. PG byte-equal:
6636 // `'1 day'` ≠ `'24 hours'` now, the format reflects it.
6637 assert_eq!(format_interval(0, 0, 0), "00:00:00");
6638 assert_eq!(format_interval(0, 1, 0), "1 day");
6639 assert_eq!(format_interval(0, -1, 0), "-1 days");
6640 assert_eq!(format_interval(0, 0, 86_400_000_000), "24:00:00");
6641 assert_eq!(format_interval(0, 0, 3_600_000_000), "01:00:00");
6642 assert_eq!(format_interval(0, 1, 9_000_000), "1 day 00:00:09");
6643 assert_eq!(format_interval(14, 0, 0), "1 year 2 mons");
6644 assert_eq!(format_interval(-1, 0, 0), "-1 mons");
6645 }
6646
6647 #[test]
6648 fn interval_format_pg_byte_equal_day_vs_24h() {
6649 // v7.37.5 β — the PG-canonical distinction `'1 day'` ≠ `'24 hours'`
6650 // is preserved in the formatter, not just the parser.
6651 assert_eq!(format_interval(0, 1, 0), "1 day");
6652 assert_eq!(format_interval(0, 0, 86_400_000_000), "24:00:00");
6653 assert_ne!(
6654 format_interval(0, 1, 0),
6655 format_interval(0, 0, 86_400_000_000),
6656 );
6657 }
6658}